file_path
stringlengths 20
207
| content
stringlengths 5
3.85M
| size
int64 5
3.85M
| lang
stringclasses 9
values | avg_line_length
float64 1.33
100
| max_line_length
int64 4
993
| alphanum_fraction
float64 0.26
0.93
|
|---|---|---|---|---|---|---|
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/__init__.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
"""Locomotion environments with velocity-tracking commands.
These environments are based on the `legged_gym` environments provided by Rudin et al.
Reference:
https://github.com/leggedrobotics/legged_gym
"""
| 336 | Python | 24.923075 | 86 | 0.764881 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/mdp/__init__.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
"""This sub-module contains the functions that are specific to the locomotion environments."""
from omni.isaac.orbit.envs.mdp import * # noqa: F401, F403
from .curriculums import * # noqa: F401, F403
from .rewards import * # noqa: F401, F403
| 370 | Python | 29.916664 | 94 | 0.732432 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/mdp/curriculums.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
"""Common functions that can be used to create curriculum for the learning environment.
The functions can be passed to the :class:`omni.isaac.orbit.managers.CurriculumTermCfg` object to enable
the curriculum introduced by the function.
"""
from __future__ import annotations
import torch
from collections.abc import Sequence
from typing import TYPE_CHECKING
from omni.isaac.orbit.assets import Articulation
from omni.isaac.orbit.managers import SceneEntityCfg
from omni.isaac.orbit.terrains import TerrainImporter
if TYPE_CHECKING:
from omni.isaac.orbit.envs import RLTaskEnv
def terrain_levels_vel(
env: RLTaskEnv, env_ids: Sequence[int], asset_cfg: SceneEntityCfg = SceneEntityCfg("robot")
) -> torch.Tensor:
"""Curriculum based on the distance the robot walked when commanded to move at a desired velocity.
This term is used to increase the difficulty of the terrain when the robot walks far enough and decrease the
difficulty when the robot walks less than half of the distance required by the commanded velocity.
.. note::
It is only possible to use this term with the terrain type ``generator``. For further information
on different terrain types, check the :class:`omni.isaac.orbit.terrains.TerrainImporter` class.
Returns:
The mean terrain level for the given environment ids.
"""
# extract the used quantities (to enable type-hinting)
asset: Articulation = env.scene[asset_cfg.name]
terrain: TerrainImporter = env.scene.terrain
command = env.command_manager.get_command("base_velocity")
# compute the distance the robot walked
distance = torch.norm(asset.data.root_pos_w[env_ids, :2] - env.scene.env_origins[env_ids, :2], dim=1)
# robots that walked far enough progress to harder terrains
move_up = distance > terrain.cfg.terrain_generator.size[0] / 2
# robots that walked less than half of their required distance go to simpler terrains
move_down = distance < torch.norm(command[env_ids, :2], dim=1) * env.max_episode_length_s * 0.5
move_down *= ~move_up
# update terrain levels
terrain.update_env_origins(env_ids, move_up, move_down)
# return the mean terrain level
return torch.mean(terrain.terrain_levels.float())
| 2,376 | Python | 41.446428 | 112 | 0.742424 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/mdp/rewards.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
from __future__ import annotations
import torch
from typing import TYPE_CHECKING
from omni.isaac.orbit.managers import SceneEntityCfg
from omni.isaac.orbit.sensors import ContactSensor
if TYPE_CHECKING:
from omni.isaac.orbit.envs import RLTaskEnv
def feet_air_time(env: RLTaskEnv, command_name: str, sensor_cfg: SceneEntityCfg, threshold: float) -> torch.Tensor:
"""Reward long steps taken by the feet using L2-kernel.
This function rewards the agent for taking steps that are longer than a threshold. This helps ensure
that the robot lifts its feet off the ground and takes steps. The reward is computed as the sum of
the time for which the feet are in the air.
If the commands are small (i.e. the agent is not supposed to take a step), then the reward is zero.
"""
# extract the used quantities (to enable type-hinting)
contact_sensor: ContactSensor = env.scene.sensors[sensor_cfg.name]
# compute the reward
first_contact = contact_sensor.compute_first_contact(env.step_dt)[:, sensor_cfg.body_ids]
last_air_time = contact_sensor.data.last_air_time[:, sensor_cfg.body_ids]
reward = torch.sum((last_air_time - threshold) * first_contact, dim=1)
# no reward for zero command
reward *= torch.norm(env.command_manager.get_command(command_name)[:, :2], dim=1) > 0.1
return reward
def feet_air_time_positive_biped(env, command_name: str, threshold: float, sensor_cfg: SceneEntityCfg) -> torch.Tensor:
"""Reward long steps taken by the feet for bipeds.
This function rewards the agent for taking steps up to a specified threshold and also keep one foot at
a time in the air.
If the commands are small (i.e. the agent is not supposed to take a step), then the reward is zero.
"""
contact_sensor: ContactSensor = env.scene.sensors[sensor_cfg.name]
# compute the reward
air_time = contact_sensor.data.current_air_time[:, sensor_cfg.body_ids]
contact_time = contact_sensor.data.current_contact_time[:, sensor_cfg.body_ids]
in_contact = contact_time > 0.0
in_mode_time = torch.where(in_contact, contact_time, air_time)
single_stance = torch.sum(in_contact.int(), dim=1) == 1
reward = torch.min(torch.where(single_stance.unsqueeze(-1), in_mode_time, 0.0), dim=1)[0]
reward = torch.clamp(reward, max=threshold)
# no reward for zero command
reward *= torch.norm(env.command_manager.get_command(command_name)[:, :2], dim=1) > 0.1
return reward
| 2,595 | Python | 43.75862 | 119 | 0.717148 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/config/__init__.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
"""Configurations for velocity-based locomotion environments."""
# We leave this file empty since we don't want to expose any configs in this package directly.
# We still need this file to import the "config" module in the parent package.
| 363 | Python | 35.399996 | 94 | 0.763085 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/config/anymal_d/rough_env_cfg.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
from omni.isaac.orbit.utils import configclass
from orbit.ext_template.tasks.locomotion.velocity.velocity_env_cfg import LocomotionVelocityRoughEnvCfg
##
# Pre-defined configs
##
from omni.isaac.orbit_assets.anymal import ANYMAL_D_CFG # isort: skip
@configclass
class AnymalDRoughEnvCfg(LocomotionVelocityRoughEnvCfg):
def __post_init__(self):
# post init of parent
super().__post_init__()
# switch robot to anymal-d
self.scene.robot = ANYMAL_D_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
@configclass
class AnymalDRoughEnvCfg_PLAY(AnymalDRoughEnvCfg):
def __post_init__(self):
# post init of parent
super().__post_init__()
# make a smaller scene for play
self.scene.num_envs = 50
self.scene.env_spacing = 2.5
# spawn the robot randomly in the grid (instead of their terrain levels)
self.scene.terrain.max_init_terrain_level = None
# reduce the number of terrains to save memory
if self.scene.terrain.terrain_generator is not None:
self.scene.terrain.terrain_generator.num_rows = 5
self.scene.terrain.terrain_generator.num_cols = 5
self.scene.terrain.terrain_generator.curriculum = False
# disable randomization for play
self.observations.policy.enable_corruption = False
# remove random pushing
self.randomization.base_external_force_torque = None
self.randomization.push_robot = None
| 1,608 | Python | 33.234042 | 103 | 0.687189 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/config/anymal_d/flat_env_cfg.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
from omni.isaac.orbit.utils import configclass
from .rough_env_cfg import AnymalDRoughEnvCfg
@configclass
class AnymalDFlatEnvCfg(AnymalDRoughEnvCfg):
def __post_init__(self):
# post init of parent
super().__post_init__()
# override rewards
self.rewards.flat_orientation_l2.weight = -5.0
self.rewards.dof_torques_l2.weight = -2.5e-5
self.rewards.feet_air_time.weight = 0.5
# change terrain to flat
self.scene.terrain.terrain_type = "plane"
self.scene.terrain.terrain_generator = None
# no height scan
self.scene.height_scanner = None
self.observations.policy.height_scan = None
# no terrain curriculum
self.curriculum.terrain_levels = None
class AnymalDFlatEnvCfg_PLAY(AnymalDFlatEnvCfg):
def __post_init__(self) -> None:
# post init of parent
super().__post_init__()
# make a smaller scene for play
self.scene.num_envs = 50
self.scene.env_spacing = 2.5
# disable randomization for play
self.observations.policy.enable_corruption = False
# remove random pushing
self.randomization.base_external_force_torque = None
self.randomization.push_robot = None
| 1,382 | Python | 30.431817 | 60 | 0.656295 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/config/anymal_d/__init__.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
import gymnasium as gym
from . import agents, flat_env_cfg, rough_env_cfg
##
# Register Gym environments.
##
gym.register(
id="Template-Velocity-Flat-Anymal-D-v0",
entry_point="omni.isaac.orbit.envs:RLTaskEnv",
disable_env_checker=True,
kwargs={
"env_cfg_entry_point": flat_env_cfg.AnymalDFlatEnvCfg,
"rsl_rl_cfg_entry_point": agents.rsl_rl_cfg.AnymalDFlatPPORunnerCfg,
},
)
gym.register(
id="Template-Velocity-Flat-Anymal-D-Play-v0",
entry_point="omni.isaac.orbit.envs:RLTaskEnv",
disable_env_checker=True,
kwargs={
"env_cfg_entry_point": flat_env_cfg.AnymalDFlatEnvCfg_PLAY,
"rsl_rl_cfg_entry_point": agents.rsl_rl_cfg.AnymalDFlatPPORunnerCfg,
},
)
gym.register(
id="Template-Velocity-Rough-Anymal-D-v0",
entry_point="omni.isaac.orbit.envs:RLTaskEnv",
disable_env_checker=True,
kwargs={
"env_cfg_entry_point": rough_env_cfg.AnymalDRoughEnvCfg,
"rsl_rl_cfg_entry_point": agents.rsl_rl_cfg.AnymalDRoughPPORunnerCfg,
},
)
gym.register(
id="Template-Velocity-Rough-Anymal-D-Play-v0",
entry_point="omni.isaac.orbit.envs:RLTaskEnv",
disable_env_checker=True,
kwargs={
"env_cfg_entry_point": rough_env_cfg.AnymalDRoughEnvCfg_PLAY,
"rsl_rl_cfg_entry_point": agents.rsl_rl_cfg.AnymalDRoughPPORunnerCfg,
},
)
| 1,474 | Python | 26.830188 | 77 | 0.685889 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/config/anymal_d/agents/rsl_rl_cfg.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
from omni.isaac.orbit.utils import configclass
from omni.isaac.orbit_tasks.utils.wrappers.rsl_rl import (
RslRlOnPolicyRunnerCfg,
RslRlPpoActorCriticCfg,
RslRlPpoAlgorithmCfg,
)
@configclass
class AnymalDRoughPPORunnerCfg(RslRlOnPolicyRunnerCfg):
num_steps_per_env = 24
max_iterations = 1500
save_interval = 50
experiment_name = "anymal_d_rough"
empirical_normalization = False
policy = RslRlPpoActorCriticCfg(
init_noise_std=1.0,
actor_hidden_dims=[512, 256, 128],
critic_hidden_dims=[512, 256, 128],
activation="elu",
)
algorithm = RslRlPpoAlgorithmCfg(
value_loss_coef=1.0,
use_clipped_value_loss=True,
clip_param=0.2,
entropy_coef=0.005,
num_learning_epochs=5,
num_mini_batches=4,
learning_rate=1.0e-3,
schedule="adaptive",
gamma=0.99,
lam=0.95,
desired_kl=0.01,
max_grad_norm=1.0,
)
@configclass
class AnymalDFlatPPORunnerCfg(AnymalDRoughPPORunnerCfg):
def __post_init__(self):
super().__post_init__()
self.max_iterations = 300
self.experiment_name = "anymal_d_flat"
self.policy.actor_hidden_dims = [128, 128, 128]
self.policy.critic_hidden_dims = [128, 128, 128]
| 1,417 | Python | 26.26923 | 58 | 0.645025 |
jonasmaximilian/orbit.test/orbit/ext_template/tasks/locomotion/velocity/config/anymal_d/agents/__init__.py | # Copyright (c) 2022-2024, The ORBIT Project Developers.
# All rights reserved.
#
# SPDX-License-Identifier: BSD-3-Clause
from . import rsl_rl_cfg # noqa: F401, F403
| 168 | Python | 23.142854 | 56 | 0.720238 |
jonasmaximilian/orbit.test/docs/CHANGELOG.rst | Changelog
---------
0.1.0 (2024-01-29)
~~~~~~~~~~~~~~~~~~
Added
^^^^^
* Created an initial template for building an extension or project based on Orbit
| 155 | reStructuredText | 13.181817 | 81 | 0.593548 |
j3soon/omni-nerf-extension/compose.yaml | version: "3"
services:
nerfstudio-renderer:
image: j3soon/nerfstudio-renderer
build:
context: nerfstudio_renderer
args:
- CUDA_VERSION=11.8.0
- CUDA_ARCHITECTURES=86
- OS_VERSION=22.04
- SERVER_PORT=10001
container_name: nerfstudio-renderer
ports:
- "10001:10001"
environment:
- DISPLAY=$DISPLAY
volumes:
- /tmp/.X11-unix:/tmp/.X11-unix
- ./nerfstudio_renderer/src:/src:ro
- ./assets:/workspace:ro
- cache:/home/user/.cache
shm_size: '6gb'
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: 1
capabilities: [gpu]
pygame-viewer:
image: j3soon/pygame-viewer
build:
context: pygame_viewer
container_name: pygame-viewer
stdin_open: true
tty: true
network_mode: host
environment:
- DISPLAY=$DISPLAY
volumes:
- /tmp/.X11-unix:/tmp/.X11-unix
- ./pygame_viewer:/src:ro
isaac-sim-viewer:
# Ref: https://github.com/j3soon/isaac-extended?tab=readme-ov-file#docker-container-with-display
image: j3soon/isaac-sim-viewer
build:
context: extension
container_name: isaac-sim-viewer
entrypoint: [bash]
stdin_open: true
tty: true
network_mode: host
environment:
- ACCEPT_EULA=Y
- PRIVACY_CONSENT=Y
- DISPLAY=$DISPLAY
volumes:
- ~/docker/isaac-sim/cache/kit:/isaac-sim/kit/cache:rw
- ~/docker/isaac-sim/cache/ov:/root/.cache/ov:rw
- ~/docker/isaac-sim/cache/pip:/root/.cache/pip:rw
- ~/docker/isaac-sim/cache/glcache:/root/.cache/nvidia/GLCache:rw
- ~/docker/isaac-sim/cache/computecache:/root/.nv/ComputeCache:rw
- ~/docker/isaac-sim/logs:/root/.nvidia-omniverse/logs:rw
- ~/docker/isaac-sim/data:/root/.local/share/ov/data:rw
- ~/docker/isaac-sim/documents:/root/Documents:rw
- /tmp/.X11-unix:/tmp/.X11-unix
- ./assets:/workspace
- ./extension:/src
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]
volumes:
cache:
| 2,191 | YAML | 27.102564 | 100 | 0.599726 |
j3soon/omni-nerf-extension/extension/README.md | # Extension Project Template
This project was automatically generated.
- `app` - It is a folder link to the location of your *Omniverse Kit* based app.
- `exts` - It is a folder where you can add new extensions. It was automatically added to extension search path. (Extension Manager -> Gear Icon -> Extension Search Path).
Open this folder using Visual Studio Code. It will suggest you to install few extensions that will make python experience better.
Look for "omni.nerf.viewport" extension in extension manager and enable it. Try applying changes to any python files, it will hot-reload and you can observe results immediately.
Alternatively, you can launch your app from console with this folder added to search path and your extension enabled, e.g.:
```
> app\omni.code.bat --ext-folder exts --enable company.hello.world
```
# App Link Setup
If `app` folder link doesn't exist or broken it can be created again. For better developer experience it is recommended to create a folder link named `app` to the *Omniverse Kit* app installed from *Omniverse Launcher*. Convenience script to use is included.
Run:
```
> link_app.bat
```
If successful you should see `app` folder link in the root of this repo.
If multiple Omniverse apps is installed script will select recommended one. Or you can explicitly pass an app:
```
> link_app.bat --app create
```
You can also just pass a path to create link to:
```
> link_app.bat --path "C:/Users/bob/AppData/Local/ov/pkg/create-2021.3.4"
```
# Sharing Your Extensions
This folder is ready to be pushed to any git repository. Once pushed direct link to a git repository can be added to *Omniverse Kit* extension search paths.
Link might look like this: `git://github.com/[user]/[your_repo].git?branch=main&dir=exts`
Notice `exts` is repo subfolder with extensions. More information can be found in "Git URL as Extension Search Paths" section of developers manual.
To add a link to your *Omniverse Kit* based app go into: Extension Manager -> Gear Icon -> Extension Search Path
| 2,042 | Markdown | 37.547169 | 258 | 0.757101 |
j3soon/omni-nerf-extension/extension/tools/scripts/link_app.py | import argparse
import json
import os
import sys
import packmanapi
import urllib3
def find_omniverse_apps():
http = urllib3.PoolManager()
try:
r = http.request("GET", "http://127.0.0.1:33480/components")
except Exception as e:
print(f"Failed retrieving apps from an Omniverse Launcher, maybe it is not installed?\nError: {e}")
sys.exit(1)
apps = {}
for x in json.loads(r.data.decode("utf-8")):
latest = x.get("installedVersions", {}).get("latest", "")
if latest:
for s in x.get("settings", []):
if s.get("version", "") == latest:
root = s.get("launch", {}).get("root", "")
apps[x["slug"]] = (x["name"], root)
break
return apps
def create_link(src, dst):
print(f"Creating a link '{src}' -> '{dst}'")
packmanapi.link(src, dst)
APP_PRIORITIES = ["code", "create", "view"]
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Create folder link to Kit App installed from Omniverse Launcher")
parser.add_argument(
"--path",
help="Path to Kit App installed from Omniverse Launcher, e.g.: 'C:/Users/bob/AppData/Local/ov/pkg/create-2021.3.4'",
required=False,
)
parser.add_argument(
"--app", help="Name of Kit App installed from Omniverse Launcher, e.g.: 'code', 'create'", required=False
)
args = parser.parse_args()
path = args.path
if not path:
print("Path is not specified, looking for Omniverse Apps...")
apps = find_omniverse_apps()
if len(apps) == 0:
print(
"Can't find any Omniverse Apps. Use Omniverse Launcher to install one. 'Code' is the recommended app for developers."
)
sys.exit(0)
print("\nFound following Omniverse Apps:")
for i, slug in enumerate(apps):
name, root = apps[slug]
print(f"{i}: {name} ({slug}) at: '{root}'")
if args.app:
selected_app = args.app.lower()
if selected_app not in apps:
choices = ", ".join(apps.keys())
print(f"Passed app: '{selected_app}' is not found. Specify one of the following found Apps: {choices}")
sys.exit(0)
else:
selected_app = next((x for x in APP_PRIORITIES if x in apps), None)
if not selected_app:
selected_app = next(iter(apps))
print(f"\nSelected app: {selected_app}")
_, path = apps[selected_app]
if not os.path.exists(path):
print(f"Provided path doesn't exist: {path}")
else:
SCRIPT_ROOT = os.path.dirname(os.path.realpath(__file__))
create_link(f"{SCRIPT_ROOT}/../../app", path)
print("Success!")
| 2,814 | Python | 32.117647 | 133 | 0.562189 |
j3soon/omni-nerf-extension/extension/tools/packman/config.packman.xml | <config remotes="cloudfront">
<remote2 name="cloudfront">
<transport actions="download" protocol="https" packageLocation="d4i3qtqj3r0z5.cloudfront.net/${name}@${version}" />
</remote2>
</config>
| 211 | XML | 34.333328 | 123 | 0.691943 |
j3soon/omni-nerf-extension/extension/tools/packman/bootstrap/install_package.py | # Copyright 2019 NVIDIA CORPORATION
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import logging
import shutil
import sys
import tempfile
import zipfile
__author__ = "hfannar"
logging.basicConfig(level=logging.WARNING, format="%(message)s")
logger = logging.getLogger("install_package")
class TemporaryDirectory:
def __init__(self):
self.path = None
def __enter__(self):
self.path = tempfile.mkdtemp()
return self.path
def __exit__(self, type, value, traceback):
# Remove temporary data created
shutil.rmtree(self.path)
def install_package(package_src_path, package_dst_path):
with zipfile.ZipFile(package_src_path, allowZip64=True) as zip_file, TemporaryDirectory() as temp_dir:
zip_file.extractall(temp_dir)
# Recursively copy (temp_dir will be automatically cleaned up on exit)
try:
# Recursive copy is needed because both package name and version folder could be missing in
# target directory:
shutil.copytree(temp_dir, package_dst_path)
except OSError as exc:
logger.warning("Directory %s already present, packaged installation aborted" % package_dst_path)
else:
logger.info("Package successfully installed to %s" % package_dst_path)
install_package(sys.argv[1], sys.argv[2])
| 1,844 | Python | 33.166666 | 108 | 0.703362 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/config/extension.toml | [package]
# Semantic Versioning is used: https://semver.org/
version = "0.0.2"
# Lists people or organizations that are considered the "authors" of the package.
authors = ["Johnson Sun", "Yan-Bin Diau"]
# The title and description fields are primarily for displaying extension info in UI
title = "omni nerf viewport"
description="A simple python extension example to use as a starting point for your extensions."
# Path (relative to the root) or content of readme markdown file for UI.
readme = "docs/README.md"
# URL of the extension source repository.
repository = ""
# One of categories for UI.
category = "Example"
# Keywords for the extension
keywords = ["kit", "example"]
# Location of change log file in target (final) folder of extension, relative to the root.
# More info on writing changelog: https://keepachangelog.com/en/1.0.0/
changelog="docs/CHANGELOG.md"
# Preview image and icon. Folder named "data" automatically goes in git lfs (see .gitattributes file).
# Preview image is shown in "Overview" of Extensions window. Screenshot of an extension might be a good preview image.
preview_image = "data/preview.png"
# Icon is shown in Extensions window, it is recommended to be square, of size 256x256.
icon = "data/icon.png"
# Use omni.ui to build simple UI
[dependencies]
"omni.kit.uiapp" = {}
# Main python module this extension provides, it will be publicly available as "import omni.nerf.viewport".
[[python.module]]
name = "omni.nerf.viewport"
[[test]]
# Extra dependencies only to be used during test run
dependencies = [
"omni.kit.ui_test" # UI testing extension
]
| 1,601 | TOML | 32.374999 | 118 | 0.74391 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/omni/nerf/viewport/extension.py | import platform
import cv2
import numpy as np
import omni.ext
import omni.ui as ui
import omni.usd
import rpyc
from omni.kit.viewport.utility import get_active_viewport
from pxr import Gf, Usd, UsdGeom
# Functions and vars are available to other extension as usual in python: `example.python_ext.some_public_function(x)`
def some_public_function(x: int):
print("[omni.nerf.viewport] some_public_function was called with x: ", x)
return x ** x
# Any class derived from `omni.ext.IExt` in top level module (defined in `python.modules` of `extension.toml`) will be
# instantiated when extension gets enabled and `on_startup(ext_id)` will be called. Later when extension gets disabled
# on_shutdown() is called.
class OmniNerfViewportExtension(omni.ext.IExt):
def __init__(self):
super().__init__()
self.is_python_supported: bool = platform.python_version().startswith("3.10")
"""The Python version must match the backend version for RPyC to work."""
self.camera_position: Gf.Vec3d = None
self.camera_rotation: Gf.Vec3d = None
# ext_id is current extension id. It can be used with extension manager to query additional information, like where
# this extension is located on filesystem.
def on_startup(self, ext_id):
# To see the Python print output in Omniverse Code, open the `Script Editor`.
# In Isaac Sim, see the startup console instead.
print("[omni.nerf.viewport] omni nerf viewport startup")
# Ref: https://docs.omniverse.nvidia.com/dev-guide/latest/programmer_ref/usd/stage/get-current-stage.html
self.usd_context = omni.usd.get_context()
# Subscribe to event streams
# Ref: https://docs.omniverse.nvidia.com/kit/docs/kit-manual/latest/guide/event_streams.html
# Ref: https://docs.omniverse.nvidia.com/kit/docs/kit-manual/104.0/carb.events/carb.events.IEventStream.html#carb.events.IEventStream.create_subscription_to_pop_by_type
# Listen to rendering events. Only triggered when the viewport is rendering is updated.
# Will not be triggered when no viewport is visible on the screen.
# Examples on using `get_rendering_event_stream` can be found by installing Isaac Sim
# and searching for `get_rendering_event_stream` under `~/.local/share/ov/pkg/isaac_sim-2023.1.1`.
self.rendering_event_stream = self.usd_context.get_rendering_event_stream()
self.rendering_event_delegate = self.rendering_event_stream.create_subscription_to_pop(
self._on_rendering_event, name="NeRF Viewport Update"
)
# TODO: Consider subscribing to update events
# Ref: https://docs.omniverse.nvidia.com/dev-guide/latest/programmer_ref/events.html#subscribe-to-update-events
# Allocate memory
self.rgba_w, self.rgba_h = 1280, 720 # Follow default camera resolution 1280x720
self.rgba = np.ones((self.rgba_h, self.rgba_w, 4), dtype=np.uint8) * 128
"""RGBA image buffer. The shape is (H, W, 4), following the NumPy convention."""
self.rgba[:,:,3] = 255
# Init RPyC connection
if self.is_python_supported:
self.init_rpyc()
# Build UI
self.build_ui()
def init_rpyc(self):
# TODO: Make the following configurable
host = 'localhost'
port = 10001
model_config_path = '/workspace/outputs/poster/nerfacto/DATE_TIME/config.yml'
model_checkpoint_path = '/workspace/outputs/poster/nerfacto/DATE_TIME/nerfstudio_models/CHECKPOINT_NAME.ckpt'
device = 'cuda'
self.rpyc_conn = rpyc.classic.connect(host, port)
self.rpyc_conn.execute('from nerfstudio_renderer import NerfStudioRenderQueue')
self.rpyc_conn.execute('from pathlib import Path')
self.rpyc_conn.execute('import torch')
self.rpyc_conn.execute(f'rq = NerfStudioRenderQueue(model_config_path=Path("{model_config_path}"), checkpoint_path="{model_checkpoint_path}", device=torch.device("{device}"))')
def build_ui(self):
"""Build the UI. Should be called upon startup."""
# Please refer to the `Omni::UI Doc` tab in Omniverse Code for efficient development.
# Ref: https://youtu.be/j1Pwi1KRkhk
# Ref: https://github.com/NVIDIA-Omniverse
# Ref: https://youtu.be/dNLFpVhBrGs
self.ui_window = ui.Window("NeRF Viewport", width=self.rgba_w, height=self.rgba_h)
with self.ui_window.frame:
with ui.ZStack():
# Camera Viewport
# Ref: https://docs.omniverse.nvidia.com/kit/docs/omni.kit.viewport.docs/latest/overview.html#simplest-example
# Don't create a new viewport widget as below, since the viewport widget will often flicker.
# Ref: https://docs.omniverse.nvidia.com/dev-guide/latest/release-notes/known-limits.html
# ```
# from omni.kit.widget.viewport import ViewportWidget
# self.ui_viewport_widget = ViewportWidget(
# resolution = (640, 360),
# width = 640,
# height = 360,
# )
# self.viewport_api = self.ui_viewport_widget.viewport_api
# ````
# Ref: https://docs.omniverse.nvidia.com/dev-guide/latest/python-snippets/viewport/change-viewport-active-camera.html
# Instead, the viewport is obtained from the active viewport in new renderings.
# NeRF Viewport
# Examples on using ByteImageProvider can be found by installing Isaac Sim
# and searching for `set_bytes_data` under `~/.local/share/ov/pkg/isaac_sim-2023.1.1`.
# Ref: https://docs.omniverse.nvidia.com/kit/docs/omni.ui/latest/omni.ui/omni.ui.ByteImageProvider.html
# Ref: https://docs.omniverse.nvidia.com/kit/docs/omni.ui/latest/omni.ui/omni.ui.ImageWithProvider.html
self.ui_nerf_provider = ui.ByteImageProvider()
# TODO: Potentially optimize with `set_bytes_data_from_gpu`
self.ui_nerf_img = ui.ImageWithProvider(
self.ui_nerf_provider,
width=ui.Percent(100),
height=ui.Percent(100),
)
# TODO: Larger image size?
with ui.VStack(height=0):
self.ui_lbl_py = ui.Label("(To Be Updated)")
state = "supported" if platform.python_version().startswith("3.10") else "NOT supported"
self.ui_lbl_py.text = f"Python {platform.python_version()} is {state}"
# UI for setting the NeRF mesh
# Ref: https://docs.omniverse.nvidia.com/workflows/latest/extensions/scatter_tool.html
with ui.HStack():
self.ui_lbl_mesh = ui.Label("NeRF Mesh", width=65)
# Ref: https://docs.omniverse.nvidia.com/dev-guide/latest/programmer_ref/ui/widgets/stringfield.html
self._mesh_prim_model = ui.SimpleStringModel()
ui.StringField(model=self._mesh_prim_model)
ui.Button(
" S ",
width=0,
height=0,
clicked_fn=self._on_btn_set_click,
tooltip="Get From Selection",
)
ui.Button("Reset Camera", width=20, clicked_fn=self.on_btn_reset_click)
self.update_ui()
def update_ui(self):
print("[omni.nerf.viewport] Updating UI")
# Ref: https://forums.developer.nvidia.com/t/refresh-window-ui/221200
self.ui_window.frame.rebuild()
def _on_btn_set_click(self):
self._mesh_prim_model.as_string = self._get_selected_prim_path()
def on_btn_reset_click(self):
# TODO: Allow resetting the camera to a specific position
# Below doesn't seem to work
# stage: Usd.Stage = self.usd_context.get_stage()
# prim: Usd.Prim = stage.GetPrimAtPath('/OmniverseKit_Persp')
# # `UsdGeom.Xformable(prim).SetTranslateOp` doesn't seem to exist
# prim.GetAttribute("xformOp:translate").Set(Gf.Vec3d(0, 0, 0.1722))
# prim.GetAttribute("xformOp:rotateXYZ").Set(Gf.Vec3d(0, -152, 0))
# print("translateOp", prim.GetAttribute("xformOp:translate").Get())
# print("rotateXYZOp", prim.GetAttribute("xformOp:rotateXYZ").Get())
print("[omni.nerf.viewport] (TODO) Reset Camera")
def _get_selected_prim_path(self):
"""Get the selected prim. Return '' if no prim is selected."""
# Ref: https://docs.omniverse.nvidia.com/workflows/latest/extensions/object_info.html#step-5-get-the-selected-prims-data
selected_prim_paths = self.usd_context.get_selection().get_selected_prim_paths()
if not selected_prim_paths:
return ''
return selected_prim_paths[0]
def _on_rendering_event(self, event):
"""Called by rendering_event_stream."""
# No need to check event type, since there is only one event type: `NEW_FRAME`.
if self.is_python_supported and self._mesh_prim_model.as_string != '':
viewport_api = get_active_viewport()
# We chose to use Viewport instead of Isaac Sim's Camera Sensor to avoid dependency on Isaac Sim.
# We want the extension to work with any Omniverse app, not just Isaac Sim.
# Ref: https://docs.omniverse.nvidia.com/isaacsim/latest/features/sensors_simulation/isaac_sim_sensors_camera.html
camera_to_world_mat: Gf.Matrix4d = viewport_api.transform
object_to_world_mat: Gf.Matrix4d = Gf.Matrix4d()
if self._mesh_prim_model.as_string != '':
stage: Usd.Stage = self.usd_context.get_stage()
selected_prim: Usd.Prim = stage.GetPrimAtPath(self._mesh_prim_model.as_string)
selected_xform: UsdGeom.Xformable = UsdGeom.Xformable(selected_prim)
object_to_world_mat = selected_xform.GetLocalTransformation()
# In USD, pre-multiplication is used for matrices.
# Ref: https://openusd.org/dev/api/usd_geom_page_front.html#UsdGeom_LinAlgBasics
world_to_object_mat: Gf.Matrix4d = object_to_world_mat.GetInverse()
camera_to_object_mat: Gf.Matrix4d = camera_to_world_mat * world_to_object_mat
camera_to_object_pos: Gf.Vec3d = camera_to_object_mat.ExtractTranslation()
# I suspect that the `Decompose` function will extract the rotation in the order of the input axes.
# So for EulerXYZ, we want to first extract and remove the Z rotation, then Y, then X.
# Then we reverse the order to get the XYZ rotation.
# I haven't spend time looking into the source code to confirm this hypothesis though.
# Ref: https://forums.developer.nvidia.com/t/how-to-get-euler-angle-of-the-prim-through-script-with-script-editor/269704/3
# Ref: https://github.com/PixarAnimationStudios/OpenUSD/blob/2864f3d04f396432f22ec5d6928fc37d34bb4c90/pxr/base/gf/rotation.cpp#L108
# must remove scale before rotation
camera_to_object_mat.Orthonormalize()
camera_to_object_rot: Gf.Vec3d = Gf.Vec3d(*reversed(camera_to_object_mat.ExtractRotation().Decompose(*reversed(Gf.Matrix3d()))))
# TODO: Consider using viewport camera projection matrix `viewport_api.projection`?
# Not same as below due to the potential difference in rotation matrix representation
# ```
# from scipy.spatial.transform import Rotation as R
# camera_rotation: Gf.Vec3d = R.from_matrix(camera_mat.ExtractRotationMatrix()).as_euler('xyz', degrees=True) # in degrees
# ```
# TODO: Consider object transform (if it is moved or rotated)
# No need to transform from Isaac Sim space to Nerfstudio space, since they are both in the same space.
# Ref: https://github.com/j3soon/coordinate-system-conventions
if camera_to_object_pos != self.camera_position or camera_to_object_rot != self.camera_rotation:
self.camera_position = camera_to_object_pos
self.camera_rotation = camera_to_object_rot
print("[omni.nerf.viewport] New camera position:", camera_to_object_pos)
print("[omni.nerf.viewport] New camera rotation:", camera_to_object_rot)
self.rpyc_conn.execute(f'rq.update_camera({list(camera_to_object_pos)}, {list(np.deg2rad(camera_to_object_rot))})')
image = self.rpyc_conn.eval('rq.get_rgb_image()')
if image is None:
return
print("[omni.nerf.viewport] NeRF viewport updated")
image = np.array(image) # received with shape (H*, W*, 3)
image = cv2.resize(image, (self.rgba_w, self.rgba_h), interpolation=cv2.INTER_LINEAR) # resize to (H, W, 3)
self.rgba[:,:,:3] = image * 255
else:
# If python version is not supported, render the dummy image.
self.rgba[:,:,:3] = (self.rgba[:,:,:3] + np.ones((self.rgba_h, self.rgba_w, 3), dtype=np.uint8)) % 256
self.ui_nerf_provider.set_bytes_data(self.rgba.flatten().tolist(), (self.rgba_w, self.rgba_h))
def on_shutdown(self):
print("[omni.nerf.viewport] omni nerf viewport shutdown")
if self.is_python_supported:
self.rpyc_conn.execute('del rq')
def destroy(self):
# Ref: https://docs.omniverse.nvidia.com/workflows/latest/extensions/object_info.html#step-3-4-use-usdcontext-to-listen-for-selection-changes
self.stage_event_stream = None
self.stage_event_delegate.unsubscribe()
| 13,852 | Python | 59.49345 | 184 | 0.632472 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/omni/nerf/viewport/__init__.py | from .extension import *
| 25 | Python | 11.999994 | 24 | 0.76 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/omni/nerf/viewport/tests/__init__.py | from .test_hello_world import * | 31 | Python | 30.999969 | 31 | 0.774194 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/omni/nerf/viewport/tests/test_hello_world.py | # NOTE:
# omni.kit.test - std python's unittest module with additional wrapping to add suport for async/await tests
# For most things refer to unittest docs: https://docs.python.org/3/library/unittest.html
import omni.kit.test
# Extnsion for writing UI tests (simulate UI interaction)
import omni.kit.ui_test as ui_test
# Import extension python module we are testing with absolute import path, as if we are external user (other extension)
import omni.nerf.viewport
# Having a test class dervived from omni.kit.test.AsyncTestCase declared on the root of module will make it auto-discoverable by omni.kit.test
class Test(omni.kit.test.AsyncTestCase):
# Before running each test
async def setUp(self):
pass
# After running each test
async def tearDown(self):
pass
# Actual test, notice it is "async" function, so "await" can be used if needed
async def test_hello_public_function(self):
result = omni.nerf.viewport.some_public_function(4)
self.assertEqual(result, 256)
async def test_window_button(self):
# Find a label in our window
label = ui_test.find("My Window//Frame/**/Label[*]")
# Find buttons in our window
add_button = ui_test.find("My Window//Frame/**/Button[*].text=='Add'")
reset_button = ui_test.find("My Window//Frame/**/Button[*].text=='Reset'")
# Click reset button
await reset_button.click()
self.assertEqual(label.widget.text, "empty")
await add_button.click()
self.assertEqual(label.widget.text, "count: 1")
await add_button.click()
self.assertEqual(label.widget.text, "count: 2")
| 1,672 | Python | 34.595744 | 142 | 0.681818 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/docs/CHANGELOG.md | # Changelog
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/).
## [0.0.1] - 2021-04-26
- Initial version of extension UI template with a window
| 178 | Markdown | 18.888887 | 80 | 0.702247 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/docs/README.md | # Python Extension Example [omni.nerf.viewport]
This is an example of pure python Kit extension. It is intended to be copied and serve as a template to create new extensions.
| 177 | Markdown | 34.599993 | 126 | 0.785311 |
j3soon/omni-nerf-extension/extension/exts/omni.nerf.viewport/docs/index.rst | omni.nerf.viewport
#############################
Example of Python only extension
.. toctree::
:maxdepth: 1
README
CHANGELOG
.. automodule::"omni.nerf.viewport"
:platform: Windows-x86_64, Linux-x86_64
:members:
:undoc-members:
:show-inheritance:
:imported-members:
:exclude-members: contextmanager
| 337 | reStructuredText | 15.095237 | 43 | 0.617211 |
j3soon/omni-nerf-extension/nerfstudio_renderer/README.md | # Nerfstudio Renderer
The following instructions assume you are in the `/nerfstudio_renderer` directory under the git repository root:
```sh
git clone https://github.com/j3soon/omni-nerf-extension.git
cd omni-nerf-extension
cd nerfstudio_renderer
```
## Launching NeRF Studio
> You can skip this section if you want to download the example poster model checkpoint and mesh.
Follow the [installation](https://docs.nerf.studio/quickstart/installation.html#use-docker-image) guide, specifically:
```sh
mkdir data
docker run --rm -it --gpus all \
-u $(id -u) \
-v $(pwd)/data:/workspace/ \
-v $HOME/.cache/:/home/user/.cache/ \
-p 7007:7007 \
--shm-size=12gb \
dromni/nerfstudio:0.3.4
```
The following subsections assume you have launched the container and using its interactive shell.
### Training a NeRF Model
Follow the [training model](https://docs.nerf.studio/quickstart/first_nerf.html) guide, specifically:
```sh
# in the container
# Download some test data:
ns-download-data nerfstudio --capture-name=poster
# Train model without normal prediction (used in the provided example poster assets for simplicity)
ns-train nerfacto --data data/nerfstudio/poster
# or train model with normal prediction (preferred)
ns-train nerfacto --data data/nerfstudio/poster --pipeline.model.predict-normals True
# wait for training to finish
```
> If you have trouble downloading the dataset, please refer to [this pull request](https://github.com/nerfstudio-project/nerfstudio/pull/3045).
### View the NeRF Model
```sh
# in the container
# change the DATE_TIME to the actual value
DATE_TIME=2023-12-30_111633
# View model
ns-viewer --load-config outputs/poster/nerfacto/$DATE_TIME/config.yml
# open the printed URL
```
### Exporting a Mesh
Follow the [export geometry](https://docs.nerf.studio/quickstart/export_geometry.html) guide, specifically:
```sh
# in the container
# change the DATE_TIME to the actual value
DATE_TIME=2023-12-30_111633
# Export mesh
# center is (-0.2, 0.1, -0.2)
ns-export tsdf --load-config outputs/poster/nerfacto/$DATE_TIME/config.yml --output-dir exports/mesh/ --target-num-faces 50000 --num-pixels-per-side 2048 --use-bounding-box True --bounding-box-min -0.55 -0.25 -0.55 --bounding-box-max 0.15 0.45 0.15
```
> Or use [Poisson Surface Reconstruction](https://docs.nerf.studio/quickstart/export_geometry.html#poisson-surface-reconstruction) instead, if the network supports predicting normals.
### View the Mesh
Open the mesh (`mesh.obj`) in Blender or any other 3D viewer.
## Download Model Checkpoint and Mesh
> You can skip this section if you want to train the example poster model checkpoint and extract mesh by yourself.
(TODO: Add link to a download a pre-trained model in release)
## Rename the Model Directory and the Checkpoint File
Rename the timestamp and checkpoint files to the same name as the placeholder for simplicity:
```sh
# change the DATE_TIME to the name of the placeholder
DATE_TIME=2023-12-30_111633
CHECKPOINT_NAME=step-000029999
cp -r ./data/outputs/poster/nerfacto/$DATE_TIME ./data/outputs/poster/nerfacto/DATE_TIME
mv ./data/outputs/poster/nerfacto/DATE_TIME/nerfstudio_models/$CHECKPOINT_NAME.ckpt ./data/outputs/poster/nerfacto/DATE_TIME/nerfstudio_models/CHECKPOINT_NAME.ckpt
```
You can check if the renaming succeeded with the following commands:
```sh
ls ./data/outputs/poster/nerfacto/DATE_TIME/config.yml
ls ./data/outputs/poster/nerfacto/DATE_TIME/nerfstudio_models/CHECKPOINT_NAME.ckpt
```
## Running with Docker Compose
Run the PyGame test window with the following commands:
```sh
xhost +local:docker
docker compose up
# in new shell
docker exec -it pygame-window /workspace/run.sh
# the initial execution might result in a delay due to the download of the pre-trained torch model.
# please re-run the script if the script times out.
```
> There seems to be an issue in `nerfstudio-renderer` that uses old code
> upon restart. I'm not aware of a reliable fix for this issue yet.
> However, running `docker compose down && docker rm $(docker ps -aq)`
> seems to fix the issue (`docker compose down` isn't enough). I believe
it is due to the `pip install` in docker entrypoint. Please keep this in
> mind when modifying the renderer code.
For development purposes, you can run the following command to run the
PyGame test window directly in the `nerfstudio-renderer` container:
```sh
docker compose build
xhost +local:docker
docker compose up
# in new shell
docker exec -it nerfstudio-renderer /workspace/tests/run_local.sh
```
The `run_local.sh` script will re-copy and re-install the package
before launching the PyGame window, so this method will not encounter
the old code issue mentioned above.
## Running Inside Docker
Alternatively, it is possible to connect to the server with [rpyc](https://github.com/tomerfiliba-org/rpyc) in the `pygame-window` container.
```python
import rpyc
import random
import time
# Make connection
conn = rpyc.classic.connect('localhost', port=10001)
# Imports
conn.execute('import nerfstudio_renderer')
conn.execute('from pathlib import Path')
conn.execute('import torch')
# Create a NerfStudioRenderQueue
# For some reason, netref-based methods keep resulting in timeouts.
conn.execute('rq = nerfstudio_renderer.NerfStudioRenderQueue(model_config_path=Path("/workspace/outputs/poster/nerfacto/DATE_TIME/config.yml"), checkpoint_path="/workspace/outputs/poster/nerfacto/DATE_TIME/nerfstudio_models/CHECKPOINT_NAME.ckpt", device=torch.device("cuda"))')
# Update camera pose
position = [random.random() for _ in range(3)]
rotation = [0., -152, 0.]
conn.execute(f'rq.update_camera({position}, {rotation})')
# Wait for some time...
time.sleep(3)
# Obtain a rendered image
image = conn.eval('rq.get_rgb_image()')
# Delete remote render queue
conn.execute('del rq')
```
Please note that the use of rpyc does not perfectly decouple the client and server. The client must be using the same Python version as the server, otherwise, there will be compatibility issues.
## Notes
- `NerfStudioRenderQueue.update_camera` can be called whenever needed. The renderer will progressively render better images serially. Each update to the camera will result in an asynchronous rendering series.
- `NerfStudioRenderQueue.get_rgb_image` will always return a newly rendered image.
- These two calls need not to be paired.
- After a call to `NerfStudioRenderQueue.get_rgb_image`, its return value will become `None` until:
1. Another image from a *newer* camera update is completed.
2. Another image from the same camera update is completed, in higher quality than the previous ones, and no images from newer updates have been ready at that point.
- **No-Way-Back Guarantee**: If an image from a newer update (say, the 10-th update) is ready at `NerfStudioRenderQueue.get_rgb_image` (even if it is never retrieved), it is guaranteed no image from the 1-st to 9-th updates will be given by future calls.
- Therefore, it is safe to call `NerfStudioRenderQueue.get_rgb_image` multiple times just to check if a newer render is done between these calls.
- You may not immediately get newest renders, but you will never get two renders in reversed time ordering.
| 7,197 | Markdown | 37.698925 | 277 | 0.764624 |
j3soon/omni-nerf-extension/nerfstudio_renderer/src/setup.py | from setuptools import setup
setup(
name='nerfstudio_renderer',
version='0.1',
packages=['nerfstudio_renderer'],
install_requires=['rpyc'],
)
| 159 | Python | 16.777776 | 37 | 0.666667 |
j3soon/omni-nerf-extension/nerfstudio_renderer/src/nerfstudio_renderer/renderer.py | from collections import defaultdict
from typing import Dict
import torch
import yaml
from nerfstudio.cameras.rays import RayBundle
from nerfstudio.data.scene_box import SceneBox
from nerfstudio_renderer.utils import *
class NerfStudioRenderer():
"""
The class is responsible for giving rendered images,
given the position, rotation, width, height, and fov
of a camera.
"""
def __init__(self, model_config_path, checkpoint_path, device):
"""
Parameters
----------
model_config_path : Path
The path to model configuration .yml file.
checkpoint_path : Path or str
The path to model checkpoint .ckpt file.
device : torch.device
Device for the model to run on. Usually CUDA or CPU.
"""
# Originally, `nerfstudio.utils.eval_setup` is used to load the entire Pipeline, which takes as input a TrainerConfig yml file.
# During the TrainerConfig setup (`nerfstudio.configs.base_config`) process, the constructor of VanillaPipeline is called.
# It will set up several components to form a complete pipeline, including a DataManager.
# The DataManager (VanillaDataManager) will perform operations to obtain DataParser outputs.
# During the setup process of the DataParser (NerfstudioDataParser), an assert is made, which forces the presence of training dataset.
# See: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/dataparsers/nerfstudio_dataparser.py#L86
# Thus, even when performing inference, training dataset is needed.
# The following code is a workaround that doesn't require to set up the entire Pipeline.
# It load solely the model checkpoint with regard to its TrainerConfig YAML, without needing to set up the entire Pipeline.
# Note that all code below are based on the v0.3.4 tag: https://github.com/nerfstudio-project/nerfstudio/tree/v0.3.4
self.device = device
# 1. Entrypoint `eval_setup`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L68
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L88
config = yaml.load(model_config_path.read_text(), Loader=yaml.Loader)
# Using zero or average appearance embedding is a inference-time choice,
# not a training-time choice (that would require inference-time to match such a choice).
# Therefore, we simply choose to use zero appearance embedding
# See Section B of the NeRF-W paper's supplementary material
# Ref: https://arxiv.org/abs/2008.02268v3
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/fields/nerfacto_field.py#L247-L254
if config.pipeline.model.use_average_appearance_embedding:
print("WARNING: Forcing zero appearance embedding, although model config specifies to use average appearance embedding.")
config.pipeline.model.use_average_appearance_embedding = False
# Disable predict normals
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/fields/nerfacto_field.py#L116
if config.pipeline.model.predict_normals:
print("WARNING: Forcing not predicting normals.")
config.pipeline.model.predict_normals = False
# TODO: Support configuring `eval_num_rays_per_chunk`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L92-L93
# 1.1. Call to `VanillaPipelineConfig.setup`, which inherits `InstantiateConfig.setup`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L103
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/configs/base_config.py#L52
# 1.2. Call to `VanillaPipelineConfig._target`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/configs/base_config.py#L54
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L224
# 1.3. Call to `VanillaPipeline.__init__`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L224
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L251
# 1.3.1. Call to `VanillaDataManagerConfig.setup`, which inherits `InstantiateConfig.setup`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L263-L265
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/configs/base_config.py#L54
# 1.3.2. Call to `VanillaDataManagerConfig._target`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/configs/base_config.py#L54
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/datamanagers/base_datamanager.py#L320
# 1.3.3. Call to `VanillaDataManager.__init__`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/datamanagers/base_datamanager.py#L320
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/datamanagers/base_datamanager.py#L378
# 1.3.4. Call to `get_dataparser_outputs`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/datamanagers/base_datamanager.py#L403
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/dataparsers/base_dataparser.py#L155
# 1.3.5. `_generate_dataparser_outputs`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/dataparsers/base_dataparser.py#L165
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/dataparsers/nerfstudio_dataparser.py#L85
# Gather model-related arguments
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/dataparsers/nerfstudio_dataparser.py#L256-L263
# in x,y,z order
# assumes that the scene is centered at the origin
aabb_scale = config.pipeline.datamanager.dataparser.scene_scale
scene_box = SceneBox(
aabb=torch.tensor(
[[-aabb_scale, -aabb_scale, -aabb_scale], [aabb_scale, aabb_scale, aabb_scale]], dtype=torch.float32
)
)
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/data/dataparsers/nerfstudio_dataparser.py#L319-L322
metadata={
"depth_filenames": None, # depth filenames are only required during training
"depth_unit_scale_factor": config.pipeline.datamanager.dataparser.depth_unit_scale_factor,
}
# 1.4. Call to `VanillaPipeline.setup`
# Setting num_train_data to 0 is fine, since we are not using average appearance embedding.
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L272
num_train_data = 0
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L275
grad_scaler = None # only required during training
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L270-L276
self.model = config.pipeline.model.setup(
scene_box=scene_box,
num_train_data=num_train_data,
metadata=metadata,
device=device,
grad_scaler=grad_scaler,
)
# Move model to device
self.model.to(device)
# 2. Call to `pipeline.eval()`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L105
self.model.eval()
# 3. Call to `eval_load_checkpoint`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L108
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L35
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L62
loaded_state = torch.load(checkpoint_path, map_location="cpu")
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L63
loaded_state, step = loaded_state["pipeline"], loaded_state["step"]
# 4. Call to `VanillaPipeline.load_pipeline`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/utils/eval_utils.py#L63
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L423
# Alter loaded model state dict for loading and update model to step
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L430-L433
state = {
(key[len("module.") :] if key.startswith("module.") else key): value for key, value in loaded_state.items()
}
self.model.update_to_step(step)
# 5. Call to `Pipeline.load_state_dict`
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L434
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L109
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L110-L119
is_ddp_model_state = True
model_state = {}
for key, value in state.items():
if key.startswith("_model."):
# remove the "_model." prefix from key
model_state[key[len("_model.") :]] = value
# make sure that the "module." prefix comes from DDP,
# rather than an attribute of the model named "module"
if not key.startswith("_model.module."):
is_ddp_model_state = False
# Drop the embedding layer for appearance embedding that requires the number of training images,
# since we are not using average appearance embedding.
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/fields/nerfacto_field.py#L112
model_state = { key: value for key, value in model_state.items() if 'embedding_appearance' not in key }
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L120-L122
# remove "module." prefix added by DDP
if is_ddp_model_state:
model_state = { key[len("module.") :]: value for key, value in model_state.items() }
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/pipelines/base_pipeline.py#L130
self.model.load_state_dict(model_state, strict=False)
# Ref: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/models/base_model.py#L175
@torch.no_grad()
def get_outputs_for_camera_ray_bundle(self, model, camera_ray_bundle: RayBundle, invalidated_fn) -> Dict[str, torch.Tensor]:
"""Takes in camera parameters and computes the output of the model.
Args:
camera_ray_bundle: ray bundle to calculate outputs over
"""
num_rays_per_chunk = model.config.eval_num_rays_per_chunk
image_height, image_width = camera_ray_bundle.origins.shape[:2]
num_rays = len(camera_ray_bundle)
outputs_lists = defaultdict(list)
for i in range(0, num_rays, num_rays_per_chunk):
if invalidated_fn():
return None
start_idx = i
end_idx = i + num_rays_per_chunk
ray_bundle = camera_ray_bundle.get_row_major_sliced_ray_bundle(start_idx, end_idx)
outputs = model.forward(ray_bundle=ray_bundle)
for output_name, output in outputs.items(): # type: ignore
if not torch.is_tensor(output):
# TODO: handle lists of tensors as well
continue
outputs_lists[output_name].append(output)
outputs = {}
for output_name, outputs_list in outputs_lists.items():
outputs[output_name] = torch.cat(outputs_list).view(image_height, image_width, -1) # type: ignore
return outputs
def render_at(self, position, rotation, width, height, fov, invalidated_fn):
"""
Parameters
----------
position : list[float]
A 3-element list specifying the camera position.
rotation : list[float]
A 3-element list specifying the camera rotation, in euler angles.
width : int
The width of the camera.
height : int
The height of the camera.
fov : float
The vertical field-of-view of the camera.
invalidated_fn : Callable[[], bool]
Function that returns whether the request is invalidated.
Returns
----------
np.array
An np array of rgb values.
"""
# Obtain a Cameras object, and transform it to the same device as the model.
c2w_matrix = camera_to_world_matrix(position, rotation)
cameras = create_cameras(c2w_matrix, width, height, fov).to(self.device)
# Obtain a ray bundle with this Cameras
ray_bundle = cameras.generate_rays(camera_indices=0, aabb_box=None)
# Inference
with torch.no_grad():
# See: https://github.com/nerfstudio-project/nerfstudio/blob/c87ebe34ba8b11172971ce48e44b6a8e8eb7a6fc/nerfstudio/models/base_model.py#L175
outputs = self.get_outputs_for_camera_ray_bundle(self.model, ray_bundle, invalidated_fn)
if outputs is None:
# Allow early return between the calculation of ray bundles if the request is invalidated.
return None
# Return results
return outputs['rgb'].cpu().numpy() | 15,697 | Python | 60.803149 | 172 | 0.707524 |
j3soon/omni-nerf-extension/nerfstudio_renderer/src/nerfstudio_renderer/__init__.py | from .renderer import NerfStudioRenderer
from .render_queue import RendererCameraConfig, NerfStudioRenderQueue
| 111 | Python | 36.333321 | 69 | 0.882883 |
j3soon/omni-nerf-extension/nerfstudio_renderer/src/nerfstudio_renderer/render_queue.py | import json
import threading
import time
from collections import deque
from nerfstudio_renderer.renderer import *
class RendererCameraConfig:
"""
This class contains functions used to load
camera configurations for the NerfStudioRenderQueue to use.
The configuration is a list of dicts.
The NerfStudioRenderQueue is then able to render differently
sized images with respect to each configuration,
for performance considerations for example.
"""
def __init__(self, cameras_config):
"""
Parameters
----------
cameras_config : list[dict]
A list of dicts that describes different camera configurations.
Each element is of the form {
'width': int, # The rendered image width (in pixels)
'height': int, # The rendered image height (in pixels)
'fov': float, # The vertical field-of-view of the camera
}
"""
self.cameras = cameras_config
def default_config():
"""
Returns a default configuration, where there are 3 cameras,
two for accelerated and estimated rendering, and the other for
high-resolution display.
Returns
----------
RendererCameraConfig
A default config.
"""
# These configurations are chosen empirically, and may be subject to change.
# Nerfstudio camera defaults:
# - vertical FoV: 50 degrees
# Isaac Sim camera defaults:
# - Size: 1280x720
# - Focal Length: 18.14756
# - Horizontal Aperture: 20.955
# - Vertical Aperture: (Value Unused)
# - (Calculated) horizontal FoV = math.degrees(2 * math.atan(20.955 / (2 * 18.14756))) = 60
# The following vertical FoV is chosen to follow the Isaac Sim camera defaults.
# Some useful equations:
# - focal_length = width / (2 * math.tan(math.radians(fov_horizontal) / 2))
# - focal_length = height / (2 * math.tan(math.radians(fov_vertical) / 2))
# - fov_vertical = math.degrees(2 * math.atan(height / (2 * focal_length)))
# - fov_horizontal = math.degrees(2 * math.atan(width / (2 * focal_length)))
# - fov_horizontal = math.degrees(2 * math.atan(horiz_aperture / (2 * focal_length)))
# Ref: https://forums.developer.nvidia.com/t/change-intrinsic-camera-parameters/180309/6
# - aspect_ratio = width / height
# - fov_vertical = math.degrees(2 * math.atan((height / width) * math.tan(math.radians(fov_horizontal) / 2)))
return RendererCameraConfig([
# fov_vertical = math.degrees(2 * math.atan((height / width) * math.tan(math.radians(fov_horizontal) / 2)))
# = 35.98339777135764
# 0.05x resolution
{ 'width': 64, 'height': 36, 'fov': 35.98339777135764 },
# 0.1x resolution
{ 'width': 128, 'height': 72, 'fov': 35.98339777135764 },
# 0.25x resolution
{ 'width': 320, 'height': 180, 'fov': 35.98339777135764 },
# 0.5x resolution
{ 'width': 640, 'height': 360, 'fov': 35.98339777135764 },
# 1x resolution
{ 'width': 1280, 'height': 720, 'fov': 35.98339777135764 },
])
def load_config(file_path=None):
"""
Returns a configuration defined by a json-formatted file.
Parameters
----------
file_path : str, optional
The path to the config file.
Returns
----------
RendererCameraConfig
A config specified by `file_path`, or a default one.
"""
if file_path is None:
return RendererCameraConfig.default_config()
with open(file_path, 'r') as f:
return RendererCameraConfig(json.load(f))
class NerfStudioRenderQueue():
"""
The class encapsulates NerfStudioRenderer and provides
a mechanism that aims at minimizing rendering latency,
via an interface that allows registration of rendering
requests. The render queue attempts to deliver
rendering results of the latest request in time, so
requests are not guaranteed to be served.
Attributes
----------
camera_config : RendererCameraConfig
The different configurations of cameras (different qualities, etc.).
renderer : NerfStudioRenderer
The NerfStudioRenderer used to actually give rendered images.
"""
def __init__(self,
model_config_path,
checkpoint_path,
device,
thread_count=3,
camera_config_path=None):
"""
Parameters
----------
model_config_path : str
The path to model configuration .yml file.
camera_config_path : str, optional
The path to the config file.
Uses `RendererCameraConfig.default_config()` when not assigned.
"""
# Construct camera config and renderer
self.camera_config = RendererCameraConfig.load_config(camera_config_path)
self.renderer = NerfStudioRenderer(model_config_path, checkpoint_path, device)
# Data maintained for optimization:
self._last_request_camera_position = (-np.inf, -np.inf, -np.inf)
"""The camera position of the last accepted request."""
self._last_request_camera_rotation = (-np.inf, -np.inf, -np.inf)
"""The camera rotation of the last accepted request."""
self._request_deque = deque(maxlen=thread_count)
"""The queue/buffer of render requests. Since we want to drop
stale requests/responses, the max size of the deque is simply
set as the thread count. The deque acts like a request buffer
instead of a task queue, which drops older requests when full.
"""
self._request_deque_pop_lock = threading.Lock()
"""The lock for the request deque. Although deque is
thread-safe, we still need to lock it when popping the deque
while empty to create blocking behavior.
"""
self._last_request_timestamp = time.time()
"""The timestamp of the last accepted request."""
self._last_request_timestamp_lock = threading.Lock()
"""The timestamp lock for the last request timestamp."""
self._last_response_timestamp = time.time()
"""The timestamp of the last sent response."""
self._last_response_timestamp_lock = threading.Lock()
"""The timestamp lock for the last response timestamp."""
self._image_response_buffer = None
"""The latest rendered image buffer, which will be cleared
immediately after retrieval."""
self._image_response_buffer_lock = threading.Lock()
"""The image lock for the image response buffer."""
for i in range(thread_count):
t = threading.Thread(target=self._render_task)
t.daemon = True
t.start()
# We choose to use threading here instead of multiprocessing
# due to lower overhead. We are aware of the GIL, but since
# the bottleneck should lie in the rendering process, which
# is implemented in C++ by PyTorch, the GIL should be released
# during PyTorch function calls.
# Ref: https://discuss.pytorch.org/t/can-pytorch-by-pass-python-gil/55498
# After going through some documents, we conclude that switching
# to multiprocessing may not be a good idea, since the overhead
# of inter-process communication may be high, and the
# implementation is not trivial.
def get_rgb_image(self):
"""
Retrieve the most recently ready rgb image.
If no rgb images have been rendered since last call of `get_rgb_image`, returns None.
Returns
----------
np.array or None
If applicable, returns an np array of size (width, height, 3) and with values ranging from 0 to 1.
Otherwise, returns None.
"""
with self._image_response_buffer_lock:
image = self._image_response_buffer
self._image_response_buffer = None
return image
def update_camera(self, position, rotation):
"""
Notifies an update to the camera pose.
This may or may not result in a new render request.
Parameters
----------
position : list[float]
A 3-element list specifying the camera position.
rotation : list[float]
A 3-element list specifying the camera rotation, in euler angles.
"""
if self._is_input_similar(position, rotation):
return
self._last_request_camera_position = position.copy()
self._last_request_camera_rotation = rotation.copy()
now = time.time()
with self._last_request_timestamp_lock:
self._last_request_timestamp = now
# Queue this render request, with request timestamp attached.
self._request_deque.append((position, rotation, now))
def _render_task(self):
while True:
with self._request_deque_pop_lock:
if len(self._request_deque) == 0:
time.sleep(0.05)
continue
task = self._request_deque.pop()
position, rotation, timestamp = task
# For each render request, render lower quality images first, and then higher quality ones.
# This rendering request and response may be dropped, as newer requests/responses invalidate older ones.
for camera in self.camera_config.cameras:
# A request can be invalidated if there are newer requests.
with self._last_request_timestamp_lock:
if timestamp - self._last_request_timestamp < 0:
continue
# Render the image
# Early return if the request is invalidated.
def request_invalidated():
with self._last_request_timestamp_lock:
return timestamp - self._last_request_timestamp < 0
image = self.renderer.render_at(position, rotation, camera['width'], camera['height'], camera['fov'], request_invalidated)
if image is None:
continue
# A response must be dropped if there are newer responses.
with self._last_response_timestamp_lock:
if timestamp - self._last_response_timestamp < 0:
continue
self._last_response_timestamp = timestamp
with self._image_response_buffer_lock:
self._image_response_buffer = image
# Checks if camera pose is similar to what was recorded.
def _is_input_similar(self, position, rotation):
return position == self._last_request_camera_position and rotation == self._last_request_camera_rotation
| 11,091 | Python | 41.992248 | 138 | 0.600126 |
j3soon/omni-nerf-extension/nerfstudio_renderer/src/nerfstudio_renderer/utils.py | import numpy as np
import torch
from nerfstudio.cameras.cameras import Cameras, CameraType
from nerfstudio.viewer.server.utils import three_js_perspective_camera_focal_length
from scipy.spatial.transform import Rotation as R
def camera_to_world_matrix(position, rotation):
"""
Constructs a camera-to-world (c2w) transformation matrix,
based on the position and rotation of the camera.
Parameters
----------
position : list[float]
A 3-element list of floats representing the position of the camera.
rotation : list[float]
A 3-element list of floats representing the rotation of the camera, in euler angles.
Returns
----------
np.array
A 4x4 camera-to-world matrix.
"""
camera_to_world_matrix = np.eye(4)
rot_matrix = R.from_euler('xyz', rotation).as_matrix()
camera_to_world_matrix[:3, :3] = rot_matrix
camera_to_world_matrix[:3, 3] = position
return camera_to_world_matrix
def create_cameras(camera_to_world_matrix, width, height, fov):
"""
Constructs a Cameras object based on a c2w matrix,
and a camera configuration from RendererCameraConfig.
Parameters
----------
camera_to_world_matrix : np.array
A 3-element list of floats representing the position of the camera.
width : int
The width of the camera.
height : int
The height of the camera.
fov : float
The vertical field-of-view of the camera.
Returns
----------
Cameras
A Cameras object.
"""
# Compute camera focal length
focal_length = three_js_perspective_camera_focal_length(fov, height)
# Only use the first 3 rows of the c2w matrix, as the last row is always [0 0 0 1].
camera_to_worlds = torch.tensor(camera_to_world_matrix)[:3].view(1, 3, 4)
return Cameras(
fx=torch.tensor([focal_length]),
fy=torch.tensor([focal_length]),
cx=width/2,
cy=height/2,
camera_to_worlds=camera_to_worlds,
camera_type=CameraType.PERSPECTIVE,
times=None,
)
| 2,071 | Python | 28.183098 | 92 | 0.654273 |
j3soon/omni-nerf-extension/pygame_viewer/pygame_test.py | import argparse
import time
import cv2
import numpy as np
import pygame
import rpyc
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument("--host", type=str, default='localhost')
parser.add_argument("--port", type=int, default=10001)
parser.add_argument("--model_config_path", type=str, required=True)
parser.add_argument("--model_checkpoint_path", type=str, required=True)
parser.add_argument("--device", type=str, choices=['cpu', 'cuda'], default='cuda')
parser.add_argument("--rpyc", type=bool, default=False)
args = parser.parse_args()
return args
def main(args):
if not args.rpyc:
# Remote: Make Connection & Import
conn = rpyc.classic.connect(args.host, args.port)
conn.execute('from nerfstudio_renderer import NerfStudioRenderQueue')
conn.execute('from pathlib import Path')
conn.execute('import torch')
else:
from nerfstudio_renderer import NerfStudioRenderQueue
from pathlib import Path
import torch
if not args.rpyc:
# Create a Remote NerfStudioRenderQueue
conn.execute(f'rq = NerfStudioRenderQueue(model_config_path=Path("{args.model_config_path}"), checkpoint_path="{args.model_checkpoint_path}", device=torch.device("{args.device}"))')
else:
rq = NerfStudioRenderQueue(
model_config_path=Path(args.model_config_path),
checkpoint_path=args.model_checkpoint_path,
device=torch.device(args.device),
)
# Initialize Pygame
pygame.init()
# Set the width and height of the window
width, height = 640, 360
window_size = (width, height)
# Create a Pygame window
screen = pygame.display.set_mode(window_size)
# Create a clock to control the frame rate
clock = pygame.time.Clock()
# Camera curve time & global screen buffer
camera_curve_time = 0
screen_buffer = np.zeros((width, height, 3), dtype=np.uint8)
# Camera pose for the poster NeRF model
camera_position = [0, 0, 0]
camera_rotation = [0, 0, 0]
running = True
while running:
for event in pygame.event.get():
if event.type == pygame.QUIT:
running = False
# Retrieve image
if not args.rpyc:
image = conn.eval('rq.get_rgb_image()')
else:
image = rq.get_rgb_image()
if image is not None:
image = np.array(image) # received with shape (H*, W*, 3)
image = cv2.resize(image, (width, height), interpolation=cv2.INTER_LINEAR) # resize to (H, W, 3)
image = np.transpose(image, (1, 0, 2))
screen_buffer[:] = image * 255
animation_progress = (np.sin(camera_curve_time) + 1) / 2
# Cover the screen buffer with an indicator of camera position
hud_width, hud_height = 100, 50
bar_x, bar_y = 20, 24
bar_w, bar_h = 60, 2
# white background
camera_position_indicator = np.ones((hud_width, hud_height, 3)) * 255
# horizontal line
camera_position_indicator[bar_x:bar_x+bar_w, bar_y:bar_y+bar_h, :] = 0
# square indicator of current position
hud_x = round(bar_x + bar_w * animation_progress)
camera_position_indicator[hud_x-5:hud_x+5, 20:30, :] = 0
screen_buffer[width-hud_width:, height-hud_height:, :] = camera_position_indicator
# Convert the NumPy array to a Pygame surface
image_surface = pygame.surfarray.make_surface(screen_buffer)
# Blit the surface to the screen
screen.blit(image_surface, (0, 0))
pygame.display.flip()
# Control the frame rate
clock.tick(30)
# Move Camera
camera_position[2] = animation_progress
# Update Camera
if not args.rpyc:
conn.execute(f'rq.update_camera({camera_position}, {camera_rotation})')
else:
rq.update_camera(camera_position, camera_rotation)
if int(time.time()) % 5 == 0:
camera_curve_time += 1.0 / 30.0
if not args.rpyc:
# Delete remote render queue
conn.execute('del rq')
# Quit Pygame
pygame.quit()
if __name__ == '__main__':
main(parse_args())
| 4,242 | Python | 32.409449 | 189 | 0.615747 |
j3soon/OmniIsaacGymEnvs-KukaReacher/setup.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Installation script for the 'isaacgymenvs' python package."""
from __future__ import absolute_import
from __future__ import print_function
from __future__ import division
from setuptools import setup, find_packages
import os
# Minimum dependencies required prior to installation
INSTALL_REQUIRES = [
"numpy==1.23.5",
"protobuf==3.20.2",
"omegaconf==2.3.0",
"hydra-core==1.3.2",
"urllib3==1.26.16",
"psutil==5.9.3",
"rl-games==1.6.1"
]
# Installation operation
setup(
name="omniisaacgymenvs",
author="NVIDIA",
version="2023.1.0a",
description="RL environments for robot learning in NVIDIA Isaac Sim.",
keywords=["robotics", "rl"],
include_package_data=True,
install_requires=INSTALL_REQUIRES,
packages=find_packages("."),
classifiers=["Natural Language :: English", "Programming Language :: Python :: 3.7, 3.8"],
zip_safe=False,
)
# EOF
| 2,473 | Python | 36.484848 | 94 | 0.740801 |
j3soon/OmniIsaacGymEnvs-KukaReacher/README.md | # Kuka Reacher Reinforcement Learning Sim2Real Environment for Omniverse Isaac Gym/Sim
This repository adds a KukaReacher environment based on [OmniIsaacGymEnvs](https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs) (commit [cc1aab0](https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs/tree/cc1aab0f904ade860fc0761d62edb6e706ab89ec)), and plan to include Sim2Real code to control a real-world [Kuka](https://www.kuka.com/en-us/products/robotics-systems/industrial-robots) with the policy learned by reinforcement learning in Omniverse Isaac Gym/Sim.
The RL code in this branch is only tested on Linux using Isaac Sim 2023.1.0. The Sim2Real code isn't implemented yet.
This repo is compatible with the following repositories:
- [OmniIsaacGymEnvs-DofbotReacher](https://github.com/j3soon/OmniIsaacGymEnvs-DofbotReacher)
- [OmniIsaacGymEnvs-UR10Reacher](https://github.com/j3soon/OmniIsaacGymEnvs-UR10Reacher)
- [OmniIsaacGymEnvs-KukaReacher](https://github.com/j3soon/OmniIsaacGymEnvs-KukaReacher)
- [OmniIsaacGymEnvs-HiwinReacher](https://github.com/j3soon/OmniIsaacGymEnvs-HiwinReacher)
## Preview
(KukaKR120R2500Pro)
## Installation
Prerequisites:
- Before starting, please make sure your hardware and software meet the [system requirements](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/requirements.html#system-requirements).
- [Install Omniverse Isaac Sim 2023.1.0](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/install_workstation.html) (Must setup Cache and Nucleus)
- You may try out newer versions of Isaac Sim along with [their corresponding patch](https://github.com/j3soon/isaac-extended#conda-issue-on-linux), but it is not guaranteed to work.
- Double check that Nucleus is correctly installed by [following these steps](https://github.com/j3soon/isaac-extended#nucleus).
- Your computer & GPU should be able to run the Cartpole example in [OmniIsaacGymEnvs](https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs)
- (Optional) [Set up a Kuka](https://www.kuka.com/en-us/products/robotics-systems/industrial-robots) in the real world
Make sure to install Isaac Sim in the default directory and clone this repository to the home directory. Otherwise, you will encounter issues if you didn't modify the commands below accordingly.
We will use Anaconda to manage our virtual environment:
1. Clone this repository and the patches repo:
- Linux
```sh
cd ~
git clone https://github.com/j3soon/OmniIsaacGymEnvs-KukaReacher.git
git clone https://github.com/j3soon/isaac-extended.git
```
- Windows
```sh
cd %USERPROFILE%
git clone https://github.com/j3soon/OmniIsaacGymEnvs-KukaReacher.git
git clone https://github.com/j3soon/isaac-extended.git
```
2. Generate [instanceable](https://docs.omniverse.nvidia.com/isaacsim/latest/isaac_gym_tutorials/tutorial_gym_instanceable_assets.html) Kuka assets for training:
[Launch the Script Editor](https://docs.omniverse.nvidia.com/app_isaacsim/app_isaacsim/tutorial_gui_interactive_scripting.html#script-editor) in Isaac Sim. Copy the content in `omniisaacgymenvs/utils/usd_utils/create_instanceable_${ROBOT_NAME_LOWER}_from_urdf.py` and execute it inside the Script Editor window. Wait until you see the text `Done!`.
The `${ROBOT_NAME_LOWER}` should be replaced with one of the following:
- `kukakr120r2500pro`
The URDF files in `/thirdparty/*` are provided by [ROS-Industrial](https://github.com/ros-industrial/kuka_experimental/tree/melodic-devel). The details on how to download this file can be found in the commit message of [fa39cbf](https://github.com/j3soon/OmniIsaacGymEnvs-KukaReacher/commit/fa39cbf37da53a7f96f3979b0a0a1f9e9a9cd103).
3. [Download and Install Anaconda](https://www.anaconda.com/products/distribution#Downloads).
```sh
# For 64-bit Linux (x86_64/x64/amd64/intel64)
wget https://repo.anaconda.com/archive/Anaconda3-2022.10-Linux-x86_64.sh
bash Anaconda3-2022.10-Linux-x86_64.sh
```
For Windows users, make sure to use `Anaconda Prompt` instead of `Anaconda Powershell Prompt`, `Command Prompt`, or `Powershell` for the following commands.
4. Patch Isaac Sim 2023.1.0
- Linux
```sh
export ISAAC_SIM="$HOME/.local/share/ov/pkg/isaac_sim-2023.1.0"
cp $ISAAC_SIM/setup_python_env.sh $ISAAC_SIM/setup_python_env.sh.bak
cp ~/isaac-extended/isaac_sim-2023.1.0-patch/linux/setup_python_env.sh $ISAAC_SIM/setup_python_env.sh
```
- Windows
> (To be updated)
5. [Set up conda environment for Isaac Sim](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/install_python.html#advanced-running-with-anaconda)
- Linux
```sh
# conda remove --name isaac-sim --all
export ISAAC_SIM="$HOME/.local/share/ov/pkg/isaac_sim-2023.1.0"
cd $ISAAC_SIM
conda env create -f environment.yml
conda activate isaac-sim
cd ~/OmniIsaacGymEnvs-KukaReacher
pip install -e .
```
- Windows
> (To be updated)
6. Activate conda environment
- Linux
```sh
export ROBOT_NAME="KukaKR120R2500Pro"
export ROBOT_NAME_LOWER="kukakr120r2500pro"
export ISAAC_SIM="$HOME/.local/share/ov/pkg/isaac_sim-2023.1.0"
cd $ISAAC_SIM
conda activate isaac-sim
source setup_conda_env.sh
```
- Windows
```sh
set ROBOT_NAME="KukaKR120R2500Pro"
set ROBOT_NAME_LOWER="kukakr120r2500pro"
set ISAAC_SIM="%LOCALAPPDATA%\ov\pkg\isaac_sim-2023.1.0"
cd %ISAAC_SIM%
conda activate isaac-sim
call setup_conda_env.bat
```
Please note that you should execute the commands in Step 6 for every new shell.
For Windows users, replace `~` to `%USERPROFILE%` for all the following commands.
## Dummy Policy
This is a sample to make sure you have setup the environment correctly. You should see a single Kuka in Isaac Sim.
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
python omniisaacgymenvs/scripts/dummy_${ROBOT_NAME_LOWER}_policy.py task=${ROBOT_NAME}Reacher test=True num_envs=1
```
Alternatively, you can replace the dummy policy with a random policy with `omniisaacgymenvs/scripts/random_policy.py`.
## Training
You can launch the training in `headless` mode as follows:
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
python omniisaacgymenvs/scripts/rlgames_train.py task=${ROBOT_NAME}Reacher headless=True
```
The number of environments is set to 2048 by default. If your GPU has small memory, you can decrease the number of environments by changing the arguments `num_envs` as below:
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
python omniisaacgymenvs/scripts/rlgames_train.py task=${ROBOT_NAME}Reacher headless=True num_envs=2048
```
You can also skip training by downloading the pre-trained model checkpoint by:
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
wget https://github.com/j3soon/OmniIsaacGymEnvs-KukaReacher/releases/download/v1.0.0/runs.zip
unzip runs.zip
```
The learning curve of the pre-trained model:
(KukaKR120R2500Pro)
## Testing
Make sure you have stored the model checkpoints at `~/OmniIsaacGymEnvs-KukaReacher/runs`, you can check it with the following command:
```sh
ls ~/OmniIsaacGymEnvs-KukaReacher/runs/${ROBOT_NAME}Reacher/nn/
```
In order to achieve the highest rewards, you may not want to use the latest checkpoint `./runs/${ROBOT_NAME}Reacher/nn/${ROBOT_NAME}Reacher.pth`. Instead, use the checkpoint with highest rewards such as `./runs/${ROBOT_NAME}Reacher/nn/last_${ROBOT_NAME}Reacher_ep_1000_rew_XXX.pth`. You can replace `${ROBOT_NAME}Reacher.pth` with the latest checkpoint before following the steps below, or simply modify the commands below to use the latest checkpoint.
You can visualize the learned policy by the following command:
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
python omniisaacgymenvs/scripts/rlgames_train.py task=${ROBOT_NAME}Reacher test=True num_envs=512 checkpoint=./runs/${ROBOT_NAME}Reacher/nn/${ROBOT_NAME}Reacher.pth
```
Likewise, you can decrease the number of environments by modifying the parameter `num_envs=512`.
## Sim2Real
> (To be updated)
## Demo
We provide an interactable demo based on the `${ROBOT_NAME}Reacher` RL example. In this demo, you can click on any of
the Kuka in the scene to manually control the robot with your keyboard as follows:
- `Q`/`A`: Control Joint 0.
- `W`/`S`: Control Joint 1.
- `E`/`D`: Control Joint 2.
- `R`/`F`: Control Joint 3.
- `T`/`G`: Control Joint 4.
- `Y`/`H`: Control Joint 5.
- `ESC`: Unselect a selected Kuka and yields manual control
Launch this demo with the following command. Note that this demo limits the maximum number of Kuka in the scene to 128.
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
python omniisaacgymenvs/scripts/rlgames_demo.py task=${ROBOT_NAME}Reacher num_envs=64
```
## Running in Docker
If you have a [NVIDIA Enterprise subscription](https://docs.omniverse.nvidia.com/prod_nucleus/prod_nucleus/enterprise/installation/planning.html), you can run all services with Docker Compose.
For users without a subscription, you can pull the [Isaac Docker image](https://catalog.ngc.nvidia.com/orgs/nvidia/containers/isaac-sim), but should still install Omniverse Nucleus beforehand. (only Isaac itself is dockerized)
Follow [this tutorial](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/install_container.html#isaac-sim-setup-remote-headless-container) to generate your NGC API Key.
Please note that you should clone this repositories in your home directory and generate instanceable assets beforehand as mentioned in the [Installation](#installation) section.
We will now set up the docker environment.
1. Build the docker image
```sh
docker pull nvcr.io/nvidia/isaac-sim:2023.1.0-hotfix.1
docker build . -t j3soon/isaac-sim
```
2. Launch an Isaac Container in Headless mode:
```sh
scripts/run_docker_headless.sh
./runheadless.native.sh
```
Alternatively, launch an Isaac Container with GUI (The host machine should include a desktop environment):
```sh
scripts/run_docker.sh
./runapp.sh
```
3. Install this repository
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
pip install -e .
```
4. Run any command in the docker container
> Make sure to add `headless=True` if the container is launched in headless mode.
For an example, running the training script:
```sh
cd ~/OmniIsaacGymEnvs-KukaReacher
python omniisaacgymenvs/scripts/rlgames_train.py task=${ROBOT_NAME}Reacher headless=True num_envs=2048
```
You can watch the training progress with:
```sh
docker exec -it isaac-sim /bin/bash
cd ~/OmniIsaacGymEnvs-KukaReacher
tensorboard --logdir=./runs
```
## Acknowledgement
This project has been made possible through the support of [ElsaLab][elsalab], [Raccoon][rccn], and [NVIDIA AI Technology Center (NVAITC)][nvaitc].
For a complete list of contributors to the code of this repository, please visit the [contributor list](https://github.com/j3soon/OmniIsaacGymEnvs-KukaReacher/graphs/contributors).
[][elsalab]
[][rccn]
[][nvaitc]
[elsalab]: https://github.com/elsa-lab
[rccn]: http://rccn.dev/
[nvaitc]: https://github.com/NVAITC
Disclaimer: this is not an official NVIDIA product.
> **Note**: below are the original README of [OmniIsaacGymEnvs](https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs).
# Omniverse Isaac Gym Reinforcement Learning Environments for Isaac Sim
## About this repository
This repository contains Reinforcement Learning examples that can be run with the latest release of [Isaac Sim](https://docs.omniverse.nvidia.com/app_isaacsim/app_isaacsim/overview.html). RL examples are trained using PPO from [rl_games](https://github.com/Denys88/rl_games) library and examples are built on top of Isaac Sim's `omni.isaac.core` and `omni.isaac.gym` frameworks.
Please see [release notes](docs/release_notes.md) for the latest updates.
<img src="https://user-images.githubusercontent.com/34286328/171454189-6afafbff-bb61-4aac-b518-24646007cb9f.gif" width="300" height="150"/> <img src="https://user-images.githubusercontent.com/34286328/184172037-cdad9ee8-f705-466f-bbde-3caa6c7dea37.gif" width="300" height="150"/>
<img src="https://user-images.githubusercontent.com/34286328/171454182-0be1b830-bceb-4cfd-93fb-e1eb8871ec68.gif" width="300" height="150"/> <img src="https://user-images.githubusercontent.com/34286328/171454193-e027885d-1510-4ef4-b838-06b37f70c1c7.gif" width="300" height="150"/>
<img src="https://user-images.githubusercontent.com/34286328/184174894-03767aa0-936c-4bfe-bbe9-a6865f539bb4.gif" width="300" height="150"/> <img src="https://user-images.githubusercontent.com/34286328/184168200-152567a8-3354-4947-9ae0-9443a56fee4c.gif" width="300" height="150"/>
<img src="https://user-images.githubusercontent.com/34286328/184176312-df7d2727-f043-46e3-b537-48a583d321b9.gif" width="300" height="150"/> <img src="https://user-images.githubusercontent.com/34286328/184178817-9c4b6b3c-c8a2-41fb-94be-cfc8ece51d5d.gif" width="300" height="150"/>
<img src="https://user-images.githubusercontent.com/34286328/171454160-8cb6739d-162a-4c84-922d-cda04382633f.gif" width="300" height="150"/> <img src="https://user-images.githubusercontent.com/34286328/171454176-ce08f6d0-3087-4ecc-9273-7d30d8f73f6d.gif" width="300" height="150"/>
<img src="https://user-images.githubusercontent.com/34286328/184170040-3f76f761-e748-452e-b8c8-3cc1c7c8cb98.gif" width="614" height="307"/>
## Installation
Follow the Isaac Sim [documentation](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/install_workstation.html) to install the latest Isaac Sim release.
*Examples in this repository rely on features from the most recent Isaac Sim release. Please make sure to update any existing Isaac Sim build to the latest release version, 2023.1.0, to ensure examples work as expected.*
Note that the 2022.2.1 OmniIsaacGymEnvs release will no longer work with the latest Isaac Sim 2023.1.0 release. Due to a change in USD APIs, line 138 in rl_task.py is no longer valid. To run the previous OIGE release with the latest Isaac Sim release, please comment out lines 137 and 138 in rl_task.py or set `add_distant_light` to `False` in the task config file. No changes are required if running with the latest release of OmniIsaacGymEnvs.
Once installed, this repository can be used as a python module, `omniisaacgymenvs`, with the python executable provided in Isaac Sim.
To install `omniisaacgymenvs`, first clone this repository:
```bash
git clone https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs.git
```
Once cloned, locate the [python executable in Isaac Sim](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/install_python.html). By default, this should be `python.sh`. We will refer to this path as `PYTHON_PATH`.
To set a `PYTHON_PATH` variable in the terminal that links to the python executable, we can run a command that resembles the following. Make sure to update the paths to your local path.
```
For Linux: alias PYTHON_PATH=~/.local/share/ov/pkg/isaac_sim-*/python.sh
For Windows: doskey PYTHON_PATH=C:\Users\user\AppData\Local\ov\pkg\isaac_sim-*\python.bat $*
For IsaacSim Docker: alias PYTHON_PATH=/isaac-sim/python.sh
```
Install `omniisaacgymenvs` as a python module for `PYTHON_PATH`:
```bash
PYTHON_PATH -m pip install -e .
```
The following error may appear during the initial installation. This error is harmless and can be ignored.
```
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
```
### Running the examples
*Note: All commands should be executed from `OmniIsaacGymEnvs/omniisaacgymenvs`.*
To train your first policy, run:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Cartpole
```
An Isaac Sim app window should be launched. Once Isaac Sim initialization completes, the Cartpole scene will be constructed and simulation will start running automatically. The process will terminate once training finishes.
Note that by default, we show a Viewport window with rendering, which slows down training. You can choose to close the Viewport window during training for better performance. The Viewport window can be re-enabled by selecting `Window > Viewport` from the top menu bar.
To achieve maximum performance, launch training in `headless` mode as follows:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Ant headless=True
```
#### A Note on the Startup Time of the Simulation
Some of the examples could take a few minutes to load because the startup time scales based on the number of environments. The startup time will continually
be optimized in future releases.
### Extension Workflow
The extension workflow provides a simple user interface for creating and launching RL tasks. To launch Isaac Sim for the extension workflow, run:
```bash
./<isaac_sim_root>/isaac-sim.gym.sh --ext-folder </parent/directory/to/OIGE>
```
Note: `isaac_sim_root` should be located in the same directory as `python.sh`.
The UI window can be activated from `Isaac Examples > RL Examples` by navigating the top menu bar.
For more details on the extension workflow, please refer to the [documentation](docs/extension_workflow.md).
### Loading trained models // Checkpoints
Checkpoints are saved in the folder `runs/EXPERIMENT_NAME/nn` where `EXPERIMENT_NAME`
defaults to the task name, but can also be overridden via the `experiment` argument.
To load a trained checkpoint and continue training, use the `checkpoint` argument:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Ant checkpoint=runs/Ant/nn/Ant.pth
```
To load a trained checkpoint and only perform inference (no training), pass `test=True`
as an argument, along with the checkpoint name. To avoid rendering overhead, you may
also want to run with fewer environments using `num_envs=64`:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Ant checkpoint=runs/Ant/nn/Ant.pth test=True num_envs=64
```
Note that if there are special characters such as `[` or `=` in the checkpoint names,
you will need to escape them and put quotes around the string. For example,
`checkpoint="runs/Ant/nn/last_Antep\=501rew\[5981.31\].pth"`
We provide pre-trained checkpoints on the [Nucleus](https://docs.omniverse.nvidia.com/nucleus/latest/index.html) server under `Assets/Isaac/2023.1.0/Isaac/Samples/OmniIsaacGymEnvs/Checkpoints`. Run the following command
to launch inference with pre-trained checkpoint:
Localhost (To set up localhost, please refer to the [Isaac Sim installation guide](https://docs.omniverse.nvidia.com/isaacsim/latest/installation/install_workstation.html)):
```bash
PYTHON_PATH scripts/rlgames_train.py task=Ant checkpoint=omniverse://localhost/NVIDIA/Assets/Isaac/2023.1.0/Isaac/Samples/OmniIsaacGymEnvs/Checkpoints/ant.pth test=True num_envs=64
```
Production server:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Ant checkpoint=http://omniverse-content-production.s3-us-west-2.amazonaws.com/Assets/Isaac/2023.1.0/Isaac/Samples/OmniIsaacGymEnvs/Checkpoints/ant.pth test=True num_envs=64
```
When running with a pre-trained checkpoint for the first time, we will automatically download the checkpoint file to `omniisaacgymenvs/checkpoints`. For subsequent runs, we will re-use the file that has already been downloaded, and will not overwrite existing checkpoints with the same name in the `checkpoints` folder.
## Runing from Docker
Latest Isaac Sim Docker image can be found on [NGC](https://catalog.ngc.nvidia.com/orgs/nvidia/containers/isaac-sim). A utility script is provided at `docker/run_docker.sh` to help initialize this repository and launch the Isaac Sim docker container. The script can be run with:
```bash
./docker/run_docker.sh
```
Then, training can be launched from the container with:
```bash
/isaac-sim/python.sh scripts/rlgames_train.py headless=True task=Ant
```
To run the Isaac Sim docker with UI, use the following script:
```bash
./docker/run_docker_viewer.sh
```
Then, training can be launched from the container with:
```bash
/isaac-sim/python.sh scripts/rlgames_train.py task=Ant
```
To avoid re-installing OIGE each time a container is launched, we also provide a dockerfile that can be used to build an image with OIGE installed. To build the image, run:
```bash
docker build -t isaac-sim-oige -f docker/dockerfile .
```
Then, start a container with the built image:
```bash
./docker/run_dockerfile.sh
```
Then, training can be launched from the container with:
```bash
/isaac-sim/python.sh scripts/rlgames_train.py task=Ant headless=True
```
## Livestream
OmniIsaacGymEnvs supports livestream through the [Omniverse Streaming Client](https://docs.omniverse.nvidia.com/app_streaming-client/app_streaming-client/overview.html). To enable this feature, add the commandline argument `enable_livestream=True`:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Ant headless=True enable_livestream=True
```
Connect from the Omniverse Streaming Client once the SimulationApp has been created. Note that enabling livestream is equivalent to training with the viewer enabled, thus the speed of training/inferencing will decrease compared to running in headless mode.
## Training Scripts
All scripts provided in `omniisaacgymenvs/scripts` can be launched directly with `PYTHON_PATH`.
To test out a task without RL in the loop, run the random policy script with:
```bash
PYTHON_PATH scripts/random_policy.py task=Cartpole
```
This script will sample random actions from the action space and apply these actions to your task without running any RL policies. Simulation should start automatically after launching the script, and will run indefinitely until terminated.
To run a simple form of PPO from `rl_games`, use the single-threaded training script:
```bash
PYTHON_PATH scripts/rlgames_train.py task=Cartpole
```
This script creates an instance of the PPO runner in `rl_games` and automatically launches training and simulation. Once training completes (the total number of iterations have been reached), the script will exit. If running inference with `test=True checkpoint=<path/to/checkpoint>`, the script will run indefinitely until terminated. Note that this script will have limitations on interaction with the UI.
### Configuration and command line arguments
We use [Hydra](https://hydra.cc/docs/intro/) to manage the config.
Common arguments for the training scripts are:
* `task=TASK` - Selects which task to use. Any of `AllegroHand`, `Ant`, `Anymal`, `AnymalTerrain`, `BallBalance`, `Cartpole`, `CartpoleCamera`, `Crazyflie`, `FactoryTaskNutBoltPick`, `FactoryTaskNutBoltPlace`, `FactoryTaskNutBoltScrew`, `FrankaCabinet`, `FrankaDeformable`, `Humanoid`, `Ingenuity`, `Quadcopter`, `ShadowHand`, `ShadowHandOpenAI_FF`, `ShadowHandOpenAI_LSTM` (these correspond to the config for each environment in the folder `omniisaacgymenvs/cfg/task`)
* `train=TRAIN` - Selects which training config to use. Will automatically default to the correct config for the environment (ie. `<TASK>PPO`).
* `num_envs=NUM_ENVS` - Selects the number of environments to use (overriding the default number of environments set in the task config).
* `seed=SEED` - Sets a seed value for randomization, and overrides the default seed in the task config
* `pipeline=PIPELINE` - Which API pipeline to use. Defaults to `gpu`, can also set to `cpu`. When using the `gpu` pipeline, all data stays on the GPU. When using the `cpu` pipeline, simulation can run on either CPU or GPU, depending on the `sim_device` setting, but a copy of the data is always made on the CPU at every step.
* `sim_device=SIM_DEVICE` - Device used for physics simulation. Set to `gpu` (default) to use GPU and to `cpu` for CPU.
* `device_id=DEVICE_ID` - Device ID for GPU to use for simulation and task. Defaults to `0`. This parameter will only be used if simulation runs on GPU.
* `rl_device=RL_DEVICE` - Which device / ID to use for the RL algorithm. Defaults to `cuda:0`, and follows PyTorch-like device syntax.
* `multi_gpu=MULTI_GPU` - Whether to train using multiple GPUs. Defaults to `False`. Note that this option is only available with `rlgames_train.py`.
* `test=TEST`- If set to `True`, only runs inference on the policy and does not do any training.
* `checkpoint=CHECKPOINT_PATH` - Path to the checkpoint to load for training or testing.
* `headless=HEADLESS` - Whether to run in headless mode.
* `enable_livestream=ENABLE_LIVESTREAM` - Whether to enable Omniverse streaming.
* `experiment=EXPERIMENT` - Sets the name of the experiment.
* `max_iterations=MAX_ITERATIONS` - Sets how many iterations to run for. Reasonable defaults are provided for the provided environments.
* `warp=WARP` - If set to True, launch the task implemented with Warp backend (Note: not all tasks have a Warp implementation).
* `kit_app=KIT_APP` - Specifies the absolute path to the kit app file to be used.
Hydra also allows setting variables inside config files directly as command line arguments. As an example, to set the minibatch size for a rl_games training run, you can use `train.params.config.minibatch_size=64`. Similarly, variables in task configs can also be set. For example, `task.env.episodeLength=100`.
#### Hydra Notes
Default values for each of these are found in the `omniisaacgymenvs/cfg/config.yaml` file.
The way that the `task` and `train` portions of the config works are through the use of config groups.
You can learn more about how these work [here](https://hydra.cc/docs/tutorials/structured_config/config_groups/)
The actual configs for `task` are in `omniisaacgymenvs/cfg/task/<TASK>.yaml` and for `train` in `omniisaacgymenvs/cfg/train/<TASK>PPO.yaml`.
In some places in the config you will find other variables referenced (for example,
`num_actors: ${....task.env.numEnvs}`). Each `.` represents going one level up in the config hierarchy.
This is documented fully [here](https://omegaconf.readthedocs.io/en/latest/usage.html#variable-interpolation).
### Tensorboard
Tensorboard can be launched during training via the following command:
```bash
PYTHON_PATH -m tensorboard.main --logdir runs/EXPERIMENT_NAME/summaries
```
## WandB support
You can run (WandB)[https://wandb.ai/] with OmniIsaacGymEnvs by setting `wandb_activate=True` flag from the command line. You can set the group, name, entity, and project for the run by setting the `wandb_group`, `wandb_name`, `wandb_entity` and `wandb_project` arguments. Make sure you have WandB installed in the Isaac Sim Python executable with `PYTHON_PATH -m pip install wandb` before activating.
## Training with Multiple GPUs
To train with multiple GPUs, use the following command, where `--proc_per_node` represents the number of available GPUs:
```bash
PYTHON_PATH -m torch.distributed.run --nnodes=1 --nproc_per_node=2 scripts/rlgames_train.py headless=True task=Ant multi_gpu=True
```
## Multi-Node Training
To train across multiple nodes/machines, it is required to launch an individual process on each node.
For the master node, use the following command, where `--proc_per_node` represents the number of available GPUs, and `--nnodes` represents the number of nodes:
```bash
PYTHON_PATH -m torch.distributed.run --nproc_per_node=2 --nnodes=2 --node_rank=0 --rdzv_id=123 --rdzv_backend=c10d --rdzv_endpoint=localhost:5555 scripts/rlgames_train.py headless=True task=Ant multi_gpu=True
```
Note that the port (`5555`) can be replaced with any other available port.
For non-master nodes, use the following command, replacing `--node_rank` with the index of each machine:
```bash
PYTHON_PATH -m torch.distributed.run --nproc_per_node=2 --nnodes=2 --node_rank=1 --rdzv_id=123 --rdzv_backend=c10d --rdzv_endpoint=ip_of_master_machine:5555 scripts/rlgames_train.py headless=True task=Ant multi_gpu=True
```
For more details on multi-node training with PyTorch, please visit [here](https://pytorch.org/tutorials/intermediate/ddp_series_multinode.html). As mentioned in the PyTorch documentation, "multinode training is bottlenecked by inter-node communication latencies". When this latency is high, it is possible multi-node training will perform worse than running on a single node instance.
## Tasks
Source code for tasks can be found in `omniisaacgymenvs/tasks`.
Each task follows the frameworks provided in `omni.isaac.core` and `omni.isaac.gym` in Isaac Sim.
Refer to [docs/framework.md](docs/framework.md) for how to create your own tasks.
Full details on each of the tasks available can be found in the [RL examples documentation](docs/rl_examples.md).
## Demo
We provide an interactable demo based on the `AnymalTerrain` RL example. In this demo, you can click on any of
the ANYmals in the scene to go into third-person mode and manually control the robot with your keyboard as follows:
- `Up Arrow`: Forward linear velocity command
- `Down Arrow`: Backward linear velocity command
- `Left Arrow`: Leftward linear velocity command
- `Right Arrow`: Rightward linear velocity command
- `Z`: Counterclockwise yaw angular velocity command
- `X`: Clockwise yaw angular velocity command
- `C`: Toggles camera view between third-person and scene view while maintaining manual control
- `ESC`: Unselect a selected ANYmal and yields manual control
Launch this demo with the following command. Note that this demo limits the maximum number of ANYmals in the scene to 128.
```
PYTHON_PATH scripts/rlgames_demo.py task=AnymalTerrain num_envs=64 checkpoint=omniverse://localhost/NVIDIA/Assets/Isaac/2023.1.0/Isaac/Samples/OmniIsaacGymEnvs/Checkpoints/anymal_terrain.pth
```
<img src="https://user-images.githubusercontent.com/34286328/184688654-6e7899b2-5847-4184-8944-2a96b129b1ff.gif" width="600" height="300"/>
| 29,919 | Markdown | 50.586207 | 469 | 0.76804 |
j3soon/OmniIsaacGymEnvs-KukaReacher/config/extension.toml | [gym]
reloadable = true
[package]
version = "0.0.0"
category = "Simulation"
title = "Isaac Gym Envs"
description = "RL environments"
authors = ["Isaac Sim Team"]
repository = "https://gitlab-master.nvidia.com/carbon-gym/omniisaacgymenvs"
keywords = ["isaac"]
changelog = "docs/CHANGELOG.md"
readme = "docs/README.md"
icon = "data/icon.png"
writeTarget.kit = true
[dependencies]
"omni.isaac.gym" = {}
"omni.isaac.core" = {}
"omni.isaac.cloner" = {}
"omni.isaac.ml_archive" = {} # torch
[[python.module]]
name = "omniisaacgymenvs"
| 532 | TOML | 20.319999 | 75 | 0.693609 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/extension.py | # Copyright (c) 2018-2023, NVIDIA CORPORATION. All rights reserved.
#
# NVIDIA CORPORATION and its licensors retain all intellectual property
# and proprietary rights in and to this software, related documentation
# and any modifications thereto. Any use, reproduction, disclosure or
# distribution of this software and related documentation without an express
# license agreement from NVIDIA CORPORATION is strictly prohibited.
#
import asyncio
import inspect
import os
import traceback
import weakref
from abc import abstractmethod
import hydra
import omni.ext
import omni.timeline
import omni.ui as ui
import omni.usd
from hydra import compose, initialize
from omegaconf import OmegaConf
from omni.isaac.cloner import GridCloner
from omni.isaac.core.utils.extensions import disable_extension, enable_extension
from omni.isaac.core.utils.torch.maths import set_seed
from omni.isaac.core.utils.viewports import set_camera_view
from omni.isaac.core.world import World
from omniisaacgymenvs.envs.vec_env_rlgames_mt import VecEnvRLGamesMT
from omniisaacgymenvs.utils.config_utils.sim_config import SimConfig
from omniisaacgymenvs.utils.hydra_cfg.reformat import omegaconf_to_dict, print_dict
from omniisaacgymenvs.utils.rlgames.rlgames_train_mt import RLGTrainer, Trainer
from omniisaacgymenvs.utils.task_util import import_tasks, initialize_task
from omni.isaac.ui.callbacks import on_open_folder_clicked, on_open_IDE_clicked
from omni.isaac.ui.menu import make_menu_item_description
from omni.isaac.ui.ui_utils import (
btn_builder,
dropdown_builder,
get_style,
int_builder,
multi_btn_builder,
multi_cb_builder,
scrolling_frame_builder,
setup_ui_headers,
str_builder,
)
from omni.kit.menu.utils import MenuItemDescription, add_menu_items, remove_menu_items
from omni.kit.viewport.utility import get_active_viewport, get_viewport_from_window_name
from omni.kit.viewport.utility.camera_state import ViewportCameraState
from pxr import Gf
ext_instance = None
class RLExtension(omni.ext.IExt):
def on_startup(self, ext_id: str):
self._render_modes = ["Full render", "UI only", "None"]
self._env = None
self._task = None
self._ext_id = ext_id
ext_manager = omni.kit.app.get_app().get_extension_manager()
extension_path = ext_manager.get_extension_path(ext_id)
self._ext_path = os.path.dirname(extension_path) if os.path.isfile(extension_path) else extension_path
self._ext_file_path = os.path.abspath(__file__)
self._initialize_task_list()
self.start_extension(
"",
"",
"RL Examples",
"RL Examples",
"",
"A set of reinforcement learning examples.",
self._ext_file_path,
)
self._task_initialized = False
self._task_changed = False
self._is_training = False
self._render = True
self._resume = False
self._test = False
self._evaluate = False
self._checkpoint_path = ""
self._timeline = omni.timeline.get_timeline_interface()
self._viewport = get_active_viewport()
self._viewport.updates_enabled = True
global ext_instance
ext_instance = self
def _initialize_task_list(self):
self._task_map, _ = import_tasks()
self._task_list = list(self._task_map.keys())
self._task_list.sort()
self._task_list.remove("CartpoleCamera") # we cannot run camera-based training from extension workflow for now. it requires a specialized app file.
self._task_name = self._task_list[0]
self._parse_config(self._task_name)
self._update_task_file_paths(self._task_name)
def _update_task_file_paths(self, task):
self._task_file_path = os.path.abspath(inspect.getfile(self._task_map[task]))
self._task_cfg_file_path = os.path.join(os.path.dirname(self._ext_file_path), f"cfg/task/{task}.yaml")
self._train_cfg_file_path = os.path.join(os.path.dirname(self._ext_file_path), f"cfg/train/{task}PPO.yaml")
def _parse_config(self, task, num_envs=None, overrides=None):
hydra.core.global_hydra.GlobalHydra.instance().clear()
initialize(version_base=None, config_path="cfg")
overrides_list = [f"task={task}"]
if overrides is not None:
overrides_list += overrides
if num_envs is None:
self._cfg = compose(config_name="config", overrides=overrides_list)
else:
self._cfg = compose(config_name="config", overrides=overrides_list + [f"num_envs={num_envs}"])
self._cfg_dict = omegaconf_to_dict(self._cfg)
self._sim_config = SimConfig(self._cfg_dict)
def start_extension(
self,
menu_name: str,
submenu_name: str,
name: str,
title: str,
doc_link: str,
overview: str,
file_path: str,
number_of_extra_frames=1,
window_width=550,
keep_window_open=False,
):
window = ui.Workspace.get_window("Property")
if window:
window.visible = False
window = ui.Workspace.get_window("Render Settings")
if window:
window.visible = False
menu_items = [make_menu_item_description(self._ext_id, name, lambda a=weakref.proxy(self): a._menu_callback())]
if menu_name == "" or menu_name is None:
self._menu_items = menu_items
elif submenu_name == "" or submenu_name is None:
self._menu_items = [MenuItemDescription(name=menu_name, sub_menu=menu_items)]
else:
self._menu_items = [
MenuItemDescription(
name=menu_name, sub_menu=[MenuItemDescription(name=submenu_name, sub_menu=menu_items)]
)
]
add_menu_items(self._menu_items, "Isaac Examples")
self._task_dropdown = None
self._cbs = None
self._build_ui(
name=name,
title=title,
doc_link=doc_link,
overview=overview,
file_path=file_path,
number_of_extra_frames=number_of_extra_frames,
window_width=window_width,
keep_window_open=keep_window_open,
)
return
def _build_ui(
self, name, title, doc_link, overview, file_path, number_of_extra_frames, window_width, keep_window_open
):
self._window = omni.ui.Window(
name, width=window_width, height=0, visible=keep_window_open, dockPreference=ui.DockPreference.LEFT_BOTTOM
)
with self._window.frame:
self._main_stack = ui.VStack(spacing=5, height=0)
with self._main_stack:
setup_ui_headers(self._ext_id, file_path, title, doc_link, overview)
self._controls_frame = ui.CollapsableFrame(
title="World Controls",
width=ui.Fraction(1),
height=0,
collapsed=False,
style=get_style(),
horizontal_scrollbar_policy=ui.ScrollBarPolicy.SCROLLBAR_AS_NEEDED,
vertical_scrollbar_policy=ui.ScrollBarPolicy.SCROLLBAR_ALWAYS_ON,
)
with self._controls_frame:
with ui.VStack(style=get_style(), spacing=5, height=0):
with ui.HStack(style=get_style()):
with ui.VStack(style=get_style(), width=ui.Fraction(20)):
dict = {
"label": "Select Task",
"type": "dropdown",
"default_val": 0,
"items": self._task_list,
"tooltip": "Select a task",
"on_clicked_fn": self._on_task_select,
}
self._task_dropdown = dropdown_builder(**dict)
with ui.Frame(tooltip="Open Source Code"):
ui.Button(
name="IconButton",
width=20,
height=20,
clicked_fn=lambda: on_open_IDE_clicked(self._ext_path, self._task_file_path),
style=get_style()["IconButton.Image::OpenConfig"],
alignment=ui.Alignment.LEFT_CENTER,
tooltip="Open in IDE",
)
with ui.Frame(tooltip="Open Task Config"):
ui.Button(
name="IconButton",
width=20,
height=20,
clicked_fn=lambda: on_open_IDE_clicked(self._ext_path, self._task_cfg_file_path),
style=get_style()["IconButton.Image::OpenConfig"],
alignment=ui.Alignment.LEFT_CENTER,
tooltip="Open in IDE",
)
with ui.Frame(tooltip="Open Training Config"):
ui.Button(
name="IconButton",
width=20,
height=20,
clicked_fn=lambda: on_open_IDE_clicked(self._ext_path, self._train_cfg_file_path),
style=get_style()["IconButton.Image::OpenConfig"],
alignment=ui.Alignment.LEFT_CENTER,
tooltip="Open in IDE",
)
dict = {
"label": "Number of environments",
"tooltip": "Enter the number of environments to construct",
"min": 0,
"max": 8192,
"default_val": self._cfg.task.env.numEnvs,
}
self._num_envs_int = int_builder(**dict)
dict = {
"label": "Load Environment",
"type": "button",
"text": "Load",
"tooltip": "Load Environment and Task",
"on_clicked_fn": self._on_load_world,
}
self._load_env_button = btn_builder(**dict)
dict = {
"label": "Rendering Mode",
"type": "dropdown",
"default_val": 0,
"items": self._render_modes,
"tooltip": "Select a rendering mode",
"on_clicked_fn": self._on_render_mode_select,
}
self._render_dropdown = dropdown_builder(**dict)
dict = {
"label": "Configure Training",
"count": 3,
"text": ["Resume from Checkpoint", "Test", "Evaluate"],
"default_val": [False, False, False],
"tooltip": [
"",
"Resume training from checkpoint",
"Play a trained policy",
"Evaluate a policy during training",
],
"on_clicked_fn": [
self._on_resume_cb_update,
self._on_test_cb_update,
self._on_evaluate_cb_update,
],
}
self._cbs = multi_cb_builder(**dict)
dict = {
"label": "Load Checkpoint",
"tooltip": "Enter path to checkpoint file",
"on_clicked_fn": self._on_checkpoint_update,
}
self._checkpoint_str = str_builder(**dict)
dict = {
"label": "Train/Test",
"count": 2,
"text": ["Start", "Stop"],
"tooltip": [
"",
"Launch new training/inference run",
"Terminate current training/inference run",
],
"on_clicked_fn": [self._on_train, self._on_train_stop],
}
self._buttons = multi_btn_builder(**dict)
return
def create_task(self):
headless = self._cfg.headless
enable_viewport = "enable_cameras" in self._cfg.task.sim and self._cfg.task.sim.enable_cameras
self._env = VecEnvRLGamesMT(
headless=headless,
sim_device=self._cfg.device_id,
enable_livestream=self._cfg.enable_livestream,
enable_viewport=enable_viewport,
launch_simulation_app=False,
)
self._task = initialize_task(self._cfg_dict, self._env, init_sim=False)
self._task_initialized = True
def _on_task_select(self, value):
if self._task_initialized and value != self._task_name:
self._task_changed = True
self._task_initialized = False
self._task_name = value
self._parse_config(self._task_name)
self._num_envs_int.set_value(self._cfg.task.env.numEnvs)
self._update_task_file_paths(self._task_name)
def _on_render_mode_select(self, value):
if value == self._render_modes[0]:
self._viewport.updates_enabled = True
window = ui.Workspace.get_window("Viewport")
window.visible = True
if self._env:
self._env._update_viewport = True
self._env._render_mode = 0
elif value == self._render_modes[1]:
self._viewport.updates_enabled = False
window = ui.Workspace.get_window("Viewport")
window.visible = False
if self._env:
self._env._update_viewport = False
self._env._render_mode = 1
elif value == self._render_modes[2]:
self._viewport.updates_enabled = False
window = ui.Workspace.get_window("Viewport")
window.visible = False
if self._env:
self._env._update_viewport = False
self._env._render_mode = 2
def _on_render_cb_update(self, value):
self._render = value
print("updates enabled", value)
self._viewport.updates_enabled = value
if self._env:
self._env._update_viewport = value
if value:
window = ui.Workspace.get_window("Viewport")
window.visible = True
else:
window = ui.Workspace.get_window("Viewport")
window.visible = False
def _on_single_env_cb_update(self, value):
visibility = "invisible" if value else "inherited"
stage = omni.usd.get_context().get_stage()
env_root = stage.GetPrimAtPath("/World/envs")
if env_root.IsValid():
for i, p in enumerate(env_root.GetChildren()):
p.GetAttribute("visibility").Set(visibility)
if value:
stage.GetPrimAtPath("/World/envs/env_0").GetAttribute("visibility").Set("inherited")
env_pos = self._task._env_pos[0].cpu().numpy().tolist()
camera_pos = [env_pos[0] + 10, env_pos[1] + 10, 3]
camera_target = [env_pos[0], env_pos[1], env_pos[2]]
else:
camera_pos = [10, 10, 3]
camera_target = [0, 0, 0]
camera_state = ViewportCameraState("/OmniverseKit_Persp", get_active_viewport())
camera_state.set_position_world(Gf.Vec3d(*camera_pos), True)
camera_state.set_target_world(Gf.Vec3d(*camera_target), True)
def _on_test_cb_update(self, value):
self._test = value
if value is True and self._checkpoint_path.strip() == "":
self._checkpoint_str.set_value(f"runs/{self._task_name}/nn/{self._task_name}.pth")
def _on_resume_cb_update(self, value):
self._resume = value
if value is True and self._checkpoint_path.strip() == "":
self._checkpoint_str.set_value(f"runs/{self._task_name}/nn/{self._task_name}.pth")
def _on_evaluate_cb_update(self, value):
self._evaluate = value
def _on_checkpoint_update(self, value):
self._checkpoint_path = value.get_value_as_string()
async def _on_load_world_async(self, use_existing_stage):
# initialize task if not initialized
if not self._task_initialized or not omni.usd.get_context().get_stage().GetPrimAtPath("/World/envs").IsValid():
self._parse_config(task=self._task_name, num_envs=self._num_envs_int.get_value_as_int())
self.create_task()
else:
# update config
self._parse_config(task=self._task_name, num_envs=self._num_envs_int.get_value_as_int())
self._task.update_config(self._sim_config)
# clear scene
# self._env._world.scene.clear()
self._env._world._sim_params = self._sim_config.get_physics_params()
await self._env._world.initialize_simulation_context_async()
set_camera_view(eye=[10, 10, 3], target=[0, 0, 0], camera_prim_path="/OmniverseKit_Persp")
if not use_existing_stage:
# clear scene
self._env._world.scene.clear()
# clear environments added to world
omni.usd.get_context().get_stage().RemovePrim("/World/collisions")
omni.usd.get_context().get_stage().RemovePrim("/World/envs")
# create scene
await self._env._world.reset_async_set_up_scene()
# update num_envs in envs
self._env.update_task_params()
else:
self._task.initialize_views(self._env._world.scene)
def _on_load_world(self):
# stop simulation before updating stage
self._timeline.stop()
asyncio.ensure_future(self._on_load_world_async(use_existing_stage=False))
def _on_train_stop(self):
if self._task_initialized:
asyncio.ensure_future(self._env._world.stop_async())
async def _on_train_async(self, overrides=None):
try:
# initialize task if not initialized
print("task initialized:", self._task_initialized)
if not self._task_initialized:
# if this is the first launch of the extension, we do not want to re-create stage if stage already exists
use_existing_stage = False
if omni.usd.get_context().get_stage().GetPrimAtPath("/World/envs").IsValid():
use_existing_stage = True
print(use_existing_stage)
await self._on_load_world_async(use_existing_stage)
# update config
self._parse_config(task=self._task_name, num_envs=self._num_envs_int.get_value_as_int(), overrides=overrides)
sim_config = SimConfig(self._cfg_dict)
self._task.update_config(sim_config)
cfg_dict = omegaconf_to_dict(self._cfg)
# sets seed. if seed is -1 will pick a random one
self._cfg.seed = set_seed(self._cfg.seed, torch_deterministic=self._cfg.torch_deterministic)
cfg_dict["seed"] = self._cfg.seed
self._checkpoint_path = self._checkpoint_str.get_value_as_string()
if self._resume or self._test:
self._cfg.checkpoint = self._checkpoint_path
self._cfg.test = self._test
self._cfg.evaluation = self._evaluate
cfg_dict["checkpoint"] = self._cfg.checkpoint
cfg_dict["test"] = self._cfg.test
cfg_dict["evaluation"] = self._cfg.evaluation
rlg_trainer = RLGTrainer(self._cfg, cfg_dict)
if not rlg_trainer._bad_checkpoint:
trainer = Trainer(rlg_trainer, self._env)
await self._env._world.reset_async_no_set_up_scene()
self._env._render_mode = self._render_dropdown.get_item_value_model().as_int
await self._env.run(trainer)
await omni.kit.app.get_app().next_update_async()
except Exception as e:
print(traceback.format_exc())
finally:
self._is_training = False
def _on_train(self):
# stop simulation if still running
self._timeline.stop()
self._on_render_mode_select(self._render_modes[self._render_dropdown.get_item_value_model().as_int])
if not self._is_training:
self._is_training = True
asyncio.ensure_future(self._on_train_async())
return
def _menu_callback(self):
self._window.visible = not self._window.visible
return
def _on_window(self, status):
return
def on_shutdown(self):
self._extra_frames = []
if self._menu_items is not None:
self._sample_window_cleanup()
self.shutdown_cleanup()
global ext_instance
ext_instance = None
return
def shutdown_cleanup(self):
return
def _sample_window_cleanup(self):
remove_menu_items(self._menu_items, "Isaac Examples")
self._window = None
self._menu_items = None
self._buttons = None
self._load_env_button = None
self._task_dropdown = None
self._cbs = None
self._checkpoint_str = None
return
def get_instance():
return ext_instance
| 22,236 | Python | 42.262646 | 155 | 0.533189 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/__init__.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import traceback
try:
from .extension import RLExtension, get_instance
# import omniisaacgymenvs.tests
except Exception as e:
pass
# print(e)
# print(traceback.format_exc())
| 1,753 | Python | 46.405404 | 80 | 0.775242 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/envs/vec_env_rlgames_mt.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import numpy as np
import torch
from omni.isaac.gym.vec_env import TaskStopException, VecEnvMT
from .vec_env_rlgames import VecEnvRLGames
# VecEnv Wrapper for RL training
class VecEnvRLGamesMT(VecEnvRLGames, VecEnvMT):
def _parse_data(self, data):
self._obs = data["obs"]
self._rew = data["rew"].to(self._task.rl_device)
self._states = torch.clamp(data["states"], -self._task.clip_obs, self._task.clip_obs).to(self._task.rl_device)
self._resets = data["reset"].to(self._task.rl_device)
self._extras = data["extras"]
def step(self, actions):
if self._stop:
raise TaskStopException()
if self._task.randomize_actions:
actions = self._task._dr_randomizer.apply_actions_randomization(
actions=actions, reset_buf=self._task.reset_buf
)
actions = torch.clamp(actions, -self._task.clip_actions, self._task.clip_actions).to(self._task.device)
self.send_actions(actions)
data = self.get_data()
if self._task.randomize_observations:
self._obs = self._task._dr_randomizer.apply_observations_randomization(
observations=self._obs.to(self._task.rl_device), reset_buf=self._task.reset_buf
)
self._obs = torch.clamp(self._obs, -self._task.clip_obs, self._task.clip_obs).to(self._task.rl_device)
obs_dict = {}
obs_dict["obs"] = self._obs
obs_dict["states"] = self._states
return obs_dict, self._rew, self._resets, self._extras
| 3,109 | Python | 42.194444 | 118 | 0.705693 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/envs/vec_env_rlgames.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from datetime import datetime
import numpy as np
import torch
from omni.isaac.gym.vec_env import VecEnvBase
# VecEnv Wrapper for RL training
class VecEnvRLGames(VecEnvBase):
def _process_data(self):
self._obs = torch.clamp(self._obs, -self._task.clip_obs, self._task.clip_obs).to(self._task.rl_device)
self._rew = self._rew.to(self._task.rl_device)
self._states = torch.clamp(self._states, -self._task.clip_obs, self._task.clip_obs).to(self._task.rl_device)
self._resets = self._resets.to(self._task.rl_device)
self._extras = self._extras
def set_task(self, task, backend="numpy", sim_params=None, init_sim=True, rendering_dt=1.0 / 60.0) -> None:
super().set_task(task, backend, sim_params, init_sim, rendering_dt)
self.num_states = self._task.num_states
self.state_space = self._task.state_space
def step(self, actions):
if self._task.randomize_actions:
actions = self._task._dr_randomizer.apply_actions_randomization(
actions=actions, reset_buf=self._task.reset_buf
)
actions = torch.clamp(actions, -self._task.clip_actions, self._task.clip_actions).to(self._task.device)
self._task.pre_physics_step(actions)
if (self.sim_frame_count + self._task.control_frequency_inv) % self._task.rendering_interval == 0:
for _ in range(self._task.control_frequency_inv - 1):
self._world.step(render=False)
self.sim_frame_count += 1
self._world.step(render=self._render)
self.sim_frame_count += 1
else:
for _ in range(self._task.control_frequency_inv):
self._world.step(render=False)
self.sim_frame_count += 1
self._obs, self._rew, self._resets, self._extras = self._task.post_physics_step()
if self._task.randomize_observations:
self._obs = self._task._dr_randomizer.apply_observations_randomization(
observations=self._obs.to(device=self._task.rl_device), reset_buf=self._task.reset_buf
)
self._states = self._task.get_states()
self._process_data()
obs_dict = {"obs": self._obs, "states": self._states}
return obs_dict, self._rew, self._resets, self._extras
def reset(self, seed=None, options=None):
"""Resets the task and applies default zero actions to recompute observations and states."""
now = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
print(f"[{now}] Running RL reset")
self._task.reset()
actions = torch.zeros((self.num_envs, self._task.num_actions), device=self._task.rl_device)
obs_dict, _, _, _ = self.step(actions)
return obs_dict
| 4,328 | Python | 43.628866 | 116 | 0.677218 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/allegro_hand.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.allegro_hand import AllegroHand
from omniisaacgymenvs.robots.articulations.views.allegro_hand_view import AllegroHandView
from omniisaacgymenvs.tasks.shared.in_hand_manipulation import InHandManipulationTask
class AllegroHandTask(InHandManipulationTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
InHandManipulationTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self.object_type = self._task_cfg["env"]["objectType"]
assert self.object_type in ["block"]
self.obs_type = self._task_cfg["env"]["observationType"]
if not (self.obs_type in ["full_no_vel", "full"]):
raise Exception("Unknown type of observations!\nobservationType should be one of: [full_no_vel, full]")
print("Obs type:", self.obs_type)
self.num_obs_dict = {
"full_no_vel": 50,
"full": 72,
}
self.object_scale = torch.tensor([1.0, 1.0, 1.0])
self._num_observations = self.num_obs_dict[self.obs_type]
self._num_actions = 16
self._num_states = 0
InHandManipulationTask.update_config(self)
def get_starting_positions(self):
self.hand_start_translation = torch.tensor([0.0, 0.0, 0.5], device=self.device)
self.hand_start_orientation = torch.tensor([0.257551, 0.283045, 0.683330, -0.621782], device=self.device)
self.pose_dy, self.pose_dz = -0.2, 0.06
def get_hand(self):
allegro_hand = AllegroHand(
prim_path=self.default_zero_env_path + "/allegro_hand",
name="allegro_hand",
translation=self.hand_start_translation,
orientation=self.hand_start_orientation,
)
self._sim_config.apply_articulation_settings(
"allegro_hand",
get_prim_at_path(allegro_hand.prim_path),
self._sim_config.parse_actor_config("allegro_hand"),
)
allegro_hand_prim = self._stage.GetPrimAtPath(allegro_hand.prim_path)
allegro_hand.set_allegro_hand_properties(stage=self._stage, allegro_hand_prim=allegro_hand_prim)
allegro_hand.set_motor_control_mode(
stage=self._stage, allegro_hand_path=self.default_zero_env_path + "/allegro_hand"
)
def get_hand_view(self, scene):
return AllegroHandView(prim_paths_expr="/World/envs/.*/allegro_hand", name="allegro_hand_view")
def get_observations(self):
self.get_object_goal_observations()
self.hand_dof_pos = self._hands.get_joint_positions(clone=False)
self.hand_dof_vel = self._hands.get_joint_velocities(clone=False)
if self.obs_type == "full_no_vel":
self.compute_full_observations(True)
elif self.obs_type == "full":
self.compute_full_observations()
else:
print("Unkown observations type!")
observations = {self._hands.name: {"obs_buf": self.obs_buf}}
return observations
def compute_full_observations(self, no_vel=False):
if no_vel:
self.obs_buf[:, 0 : self.num_hand_dofs] = unscale(
self.hand_dof_pos, self.hand_dof_lower_limits, self.hand_dof_upper_limits
)
self.obs_buf[:, 16:19] = self.object_pos
self.obs_buf[:, 19:23] = self.object_rot
self.obs_buf[:, 23:26] = self.goal_pos
self.obs_buf[:, 26:30] = self.goal_rot
self.obs_buf[:, 30:34] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
self.obs_buf[:, 34:50] = self.actions
else:
self.obs_buf[:, 0 : self.num_hand_dofs] = unscale(
self.hand_dof_pos, self.hand_dof_lower_limits, self.hand_dof_upper_limits
)
self.obs_buf[:, self.num_hand_dofs : 2 * self.num_hand_dofs] = self.vel_obs_scale * self.hand_dof_vel
self.obs_buf[:, 32:35] = self.object_pos
self.obs_buf[:, 35:39] = self.object_rot
self.obs_buf[:, 39:42] = self.object_linvel
self.obs_buf[:, 42:45] = self.vel_obs_scale * self.object_angvel
self.obs_buf[:, 45:48] = self.goal_pos
self.obs_buf[:, 48:52] = self.goal_rot
self.obs_buf[:, 52:56] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
self.obs_buf[:, 56:72] = self.actions
| 6,329 | Python | 42.655172 | 115 | 0.658872 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/ball_balance.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.objects import DynamicSphere
from omni.isaac.core.prims import RigidPrim, RigidPrimView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage
from omni.isaac.core.utils.torch.maths import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.balance_bot import BalanceBot
from pxr import PhysxSchema
class BallBalanceTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._num_observations = 12 + 12
self._num_actions = 3
self.anchored = False
RLTask.__init__(self, name, env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._dt = self._task_cfg["sim"]["dt"]
self._table_position = torch.tensor([0, 0, 0.56])
self._ball_position = torch.tensor([0.0, 0.0, 1.0])
self._ball_radius = 0.1
self._action_speed_scale = self._task_cfg["env"]["actionSpeedScale"]
self._max_episode_length = self._task_cfg["env"]["maxEpisodeLength"]
def set_up_scene(self, scene) -> None:
self.get_balance_table()
self.add_ball()
super().set_up_scene(scene, replicate_physics=False)
self.set_up_table_anchors()
self._balance_bots = ArticulationView(
prim_paths_expr="/World/envs/.*/BalanceBot/tray", name="balance_bot_view", reset_xform_properties=False
)
scene.add(self._balance_bots)
self._balls = RigidPrimView(
prim_paths_expr="/World/envs/.*/Ball/ball", name="ball_view", reset_xform_properties=False
)
scene.add(self._balls)
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("balance_bot_view"):
scene.remove_object("balance_bot_view", registry_only=True)
if scene.object_exists("ball_view"):
scene.remove_object("ball_view", registry_only=True)
self._balance_bots = ArticulationView(
prim_paths_expr="/World/envs/.*/BalanceBot/tray", name="balance_bot_view", reset_xform_properties=False
)
scene.add(self._balance_bots)
self._balls = RigidPrimView(
prim_paths_expr="/World/envs/.*/Ball/ball", name="ball_view", reset_xform_properties=False
)
scene.add(self._balls)
def get_balance_table(self):
balance_table = BalanceBot(
prim_path=self.default_zero_env_path + "/BalanceBot", name="BalanceBot", translation=self._table_position
)
self._sim_config.apply_articulation_settings(
"table", get_prim_at_path(balance_table.prim_path), self._sim_config.parse_actor_config("table")
)
def add_ball(self):
ball = DynamicSphere(
prim_path=self.default_zero_env_path + "/Ball/ball",
translation=self._ball_position,
name="ball_0",
radius=self._ball_radius,
color=torch.tensor([0.9, 0.6, 0.2]),
)
self._sim_config.apply_articulation_settings(
"ball", get_prim_at_path(ball.prim_path), self._sim_config.parse_actor_config("ball")
)
def set_up_table_anchors(self):
from pxr import Gf
height = 0.08
stage = get_current_stage()
for i in range(self._num_envs):
base_path = f"{self.default_base_env_path}/env_{i}/BalanceBot"
for j, leg_offset in enumerate([(0.4, 0, height), (-0.2, 0.34641, 0), (-0.2, -0.34641, 0)]):
# fix the legs to ground
leg_path = f"{base_path}/lower_leg{j}"
ground_joint_path = leg_path + "_ground"
env_pos = stage.GetPrimAtPath(f"{self.default_base_env_path}/env_{i}").GetAttribute("xformOp:translate").Get()
anchor_pos = env_pos + Gf.Vec3d(*leg_offset)
self.fix_to_ground(stage, ground_joint_path, leg_path, anchor_pos)
def fix_to_ground(self, stage, joint_path, prim_path, anchor_pos):
from pxr import UsdPhysics, Gf
# D6 fixed joint
d6FixedJoint = UsdPhysics.Joint.Define(stage, joint_path)
d6FixedJoint.CreateBody0Rel().SetTargets(["/World/defaultGroundPlane"])
d6FixedJoint.CreateBody1Rel().SetTargets([prim_path])
d6FixedJoint.CreateLocalPos0Attr().Set(anchor_pos)
d6FixedJoint.CreateLocalRot0Attr().Set(Gf.Quatf(1.0, Gf.Vec3f(0, 0, 0)))
d6FixedJoint.CreateLocalPos1Attr().Set(Gf.Vec3f(0, 0, 0.18))
d6FixedJoint.CreateLocalRot1Attr().Set(Gf.Quatf(1.0, Gf.Vec3f(0, 0, 0)))
# lock all DOF (lock - low is greater than high)
d6Prim = stage.GetPrimAtPath(joint_path)
limitAPI = UsdPhysics.LimitAPI.Apply(d6Prim, "transX")
limitAPI.CreateLowAttr(1.0)
limitAPI.CreateHighAttr(-1.0)
limitAPI = UsdPhysics.LimitAPI.Apply(d6Prim, "transY")
limitAPI.CreateLowAttr(1.0)
limitAPI.CreateHighAttr(-1.0)
limitAPI = UsdPhysics.LimitAPI.Apply(d6Prim, "transZ")
limitAPI.CreateLowAttr(1.0)
limitAPI.CreateHighAttr(-1.0)
def get_observations(self) -> dict:
ball_positions, ball_orientations = self._balls.get_world_poses(clone=False)
ball_positions = ball_positions[:, 0:3] - self._env_pos
ball_velocities = self._balls.get_velocities(clone=False)
ball_linvels = ball_velocities[:, 0:3]
ball_angvels = ball_velocities[:, 3:6]
dof_pos = self._balance_bots.get_joint_positions(clone=False)
dof_vel = self._balance_bots.get_joint_velocities(clone=False)
sensor_force_torques = self._balance_bots.get_measured_joint_forces(joint_indices=self._sensor_indices) # (num_envs, num_sensors, 6)
self.obs_buf[..., 0:3] = dof_pos[..., self.actuated_dof_indices]
self.obs_buf[..., 3:6] = dof_vel[..., self.actuated_dof_indices]
self.obs_buf[..., 6:9] = ball_positions
self.obs_buf[..., 9:12] = ball_linvels
self.obs_buf[..., 12:15] = sensor_force_torques[..., 0] / 20.0
self.obs_buf[..., 15:18] = sensor_force_torques[..., 3] / 20.0
self.obs_buf[..., 18:21] = sensor_force_torques[..., 4] / 20.0
self.obs_buf[..., 21:24] = sensor_force_torques[..., 5] / 20.0
self.ball_positions = ball_positions
self.ball_linvels = ball_linvels
observations = {"ball_balance": {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
# update position targets from actions
self.dof_position_targets[..., self.actuated_dof_indices] += (
self._dt * self._action_speed_scale * actions.to(self.device)
)
self.dof_position_targets[:] = tensor_clamp(
self.dof_position_targets, self.bbot_dof_lower_limits, self.bbot_dof_upper_limits
)
# reset position targets for reset envs
self.dof_position_targets[reset_env_ids] = 0
self._balance_bots.set_joint_position_targets(self.dof_position_targets) # .clone())
def reset_idx(self, env_ids):
num_resets = len(env_ids)
env_ids_32 = env_ids.type(torch.int32)
env_ids_64 = env_ids.type(torch.int64)
min_d = 0.001 # min horizontal dist from origin
max_d = 0.4 # max horizontal dist from origin
min_height = 1.0
max_height = 2.0
min_horizontal_speed = 0
max_horizontal_speed = 2
dists = torch_rand_float(min_d, max_d, (num_resets, 1), self._device)
dirs = torch_random_dir_2((num_resets, 1), self._device)
hpos = dists * dirs
speedscales = (dists - min_d) / (max_d - min_d)
hspeeds = torch_rand_float(min_horizontal_speed, max_horizontal_speed, (num_resets, 1), self._device)
hvels = -speedscales * hspeeds * dirs
vspeeds = -torch_rand_float(5.0, 5.0, (num_resets, 1), self._device).squeeze()
ball_pos = self.initial_ball_pos.clone()
ball_rot = self.initial_ball_rot.clone()
# position
ball_pos[env_ids_64, 0:2] += hpos[..., 0:2]
ball_pos[env_ids_64, 2] += torch_rand_float(min_height, max_height, (num_resets, 1), self._device).squeeze()
# rotation
ball_rot[env_ids_64, 0] = 1
ball_rot[env_ids_64, 1:] = 0
ball_velocities = self.initial_ball_velocities.clone()
# linear
ball_velocities[env_ids_64, 0:2] = hvels[..., 0:2]
ball_velocities[env_ids_64, 2] = vspeeds
# angular
ball_velocities[env_ids_64, 3:6] = 0
# reset root state for bbots and balls in selected envs
self._balls.set_world_poses(ball_pos[env_ids_64], ball_rot[env_ids_64], indices=env_ids_32)
self._balls.set_velocities(ball_velocities[env_ids_64], indices=env_ids_32)
# reset root pose and velocity
self._balance_bots.set_world_poses(
self.initial_bot_pos[env_ids_64].clone(), self.initial_bot_rot[env_ids_64].clone(), indices=env_ids_32
)
self._balance_bots.set_velocities(self.initial_bot_velocities[env_ids_64].clone(), indices=env_ids_32)
# reset DOF states for bbots in selected envs
self._balance_bots.set_joint_positions(self.initial_dof_positions[env_ids_64].clone(), indices=env_ids_32)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def post_reset(self):
dof_limits = self._balance_bots.get_dof_limits()
self.bbot_dof_lower_limits, self.bbot_dof_upper_limits = torch.t(dof_limits[0].to(device=self._device))
self.initial_dof_positions = self._balance_bots.get_joint_positions()
self.initial_bot_pos, self.initial_bot_rot = self._balance_bots.get_world_poses()
# self.initial_bot_pos[..., 2] = 0.559 # tray_height
self.initial_bot_velocities = self._balance_bots.get_velocities()
self.initial_ball_pos, self.initial_ball_rot = self._balls.get_world_poses()
self.initial_ball_velocities = self._balls.get_velocities()
self.dof_position_targets = torch.zeros(
(self.num_envs, self._balance_bots.num_dof), dtype=torch.float32, device=self._device, requires_grad=False
)
actuated_joints = ["lower_leg0", "lower_leg1", "lower_leg2"]
self.actuated_dof_indices = torch.tensor(
[self._balance_bots._dof_indices[j] for j in actuated_joints], device=self._device, dtype=torch.long
)
force_links = ["upper_leg0", "upper_leg1", "upper_leg2"]
self._sensor_indices = torch.tensor(
[self._balance_bots._body_indices[j] for j in force_links], device=self._device, dtype=torch.long
)
def calculate_metrics(self) -> None:
ball_dist = torch.sqrt(
self.ball_positions[..., 0] * self.ball_positions[..., 0]
+ (self.ball_positions[..., 2] - 0.7) * (self.ball_positions[..., 2] - 0.7)
+ (self.ball_positions[..., 1]) * self.ball_positions[..., 1]
)
ball_speed = torch.sqrt(
self.ball_linvels[..., 0] * self.ball_linvels[..., 0]
+ self.ball_linvels[..., 1] * self.ball_linvels[..., 1]
+ self.ball_linvels[..., 2] * self.ball_linvels[..., 2]
)
pos_reward = 1.0 / (1.0 + ball_dist)
speed_reward = 1.0 / (1.0 + ball_speed)
self.rew_buf[:] = pos_reward * speed_reward
def is_done(self) -> None:
reset = torch.where(
self.progress_buf >= self._max_episode_length - 1, torch.ones_like(self.reset_buf), self.reset_buf
)
reset = torch.where(
self.ball_positions[..., 2] < self._ball_radius * 1.5, torch.ones_like(self.reset_buf), reset
)
self.reset_buf[:] = reset
| 13,958 | Python | 44.174757 | 140 | 0.630391 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/cartpole_camera.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from gym import spaces
import numpy as np
import torch
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.tasks.cartpole import CartpoleTask
from omniisaacgymenvs.robots.articulations.cartpole import Cartpole
class CartpoleCameraTask(CartpoleTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._max_episode_length = 500
self._num_observations = 4
self._num_actions = 1
# use multi-dimensional observation for camera RGB
self.observation_space = spaces.Box(
np.ones((self.camera_width, self.camera_height, 3), dtype=np.float32) * -np.Inf,
np.ones((self.camera_width, self.camera_height, 3), dtype=np.float32) * np.Inf)
RLTask.__init__(self, name, env)
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._cartpole_positions = torch.tensor([0.0, 0.0, 2.0])
self._reset_dist = self._task_cfg["env"]["resetDist"]
self._max_push_effort = self._task_cfg["env"]["maxEffort"]
self.camera_type = self._task_cfg["env"].get("cameraType", 'rgb')
self.camera_width = self._task_cfg["env"]["cameraWidth"]
self.camera_height = self._task_cfg["env"]["cameraHeight"]
self.camera_channels = 3
self._export_images = self._task_cfg["env"]["exportImages"]
def cleanup(self) -> None:
# initialize remaining buffers
RLTask.cleanup(self)
# override observation buffer for camera data
self.obs_buf = torch.zeros(
(self.num_envs, self.camera_width, self.camera_height, 3), device=self.device, dtype=torch.float)
def set_up_scene(self, scene) -> None:
self.get_cartpole()
RLTask.set_up_scene(self, scene)
# start replicator to capture image data
self.rep.orchestrator._orchestrator._is_started = True
# set up cameras
self.render_products = []
env_pos = self._env_pos.cpu()
for i in range(self._num_envs):
camera = self.rep.create.camera(
position=(-4.2 + env_pos[i][0], env_pos[i][1], 3.0), look_at=(env_pos[i][0], env_pos[i][1], 2.55))
render_product = self.rep.create.render_product(camera, resolution=(self.camera_width, self.camera_height))
self.render_products.append(render_product)
# initialize pytorch writer for vectorized collection
self.pytorch_listener = self.PytorchListener()
self.pytorch_writer = self.rep.WriterRegistry.get("PytorchWriter")
self.pytorch_writer.initialize(listener=self.pytorch_listener, device="cuda")
self.pytorch_writer.attach(self.render_products)
self._cartpoles = ArticulationView(
prim_paths_expr="/World/envs/.*/Cartpole", name="cartpole_view", reset_xform_properties=False
)
scene.add(self._cartpoles)
return
def get_observations(self) -> dict:
dof_pos = self._cartpoles.get_joint_positions(clone=False)
dof_vel = self._cartpoles.get_joint_velocities(clone=False)
self.cart_pos = dof_pos[:, self._cart_dof_idx]
self.cart_vel = dof_vel[:, self._cart_dof_idx]
self.pole_pos = dof_pos[:, self._pole_dof_idx]
self.pole_vel = dof_vel[:, self._pole_dof_idx]
# retrieve RGB data from all render products
images = self.pytorch_listener.get_rgb_data()
if images is not None:
if self._export_images:
from torchvision.utils import save_image, make_grid
img = images/255
save_image(make_grid(img, nrows = 2), 'cartpole_export.png')
self.obs_buf = torch.swapaxes(images, 1, 3).clone().float()/255.0
else:
print("Image tensor is NONE!")
return self.obs_buf
| 5,824 | Python | 41.518248 | 119 | 0.67342 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/anymal_terrain.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.simulation_context import SimulationContext
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage
from omni.isaac.core.utils.torch.rotations import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.anymal import Anymal
from omniisaacgymenvs.robots.articulations.views.anymal_view import AnymalView
from omniisaacgymenvs.tasks.utils.anymal_terrain_generator import *
from omniisaacgymenvs.utils.terrain_utils.terrain_utils import *
from pxr import UsdLux, UsdPhysics
class AnymalTerrainTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.height_samples = None
self.custom_origins = False
self.init_done = False
self._env_spacing = 0.0
self._num_observations = 188
self._num_actions = 12
self.update_config(sim_config)
RLTask.__init__(self, name, env)
self.height_points = self.init_height_points()
self.measured_heights = None
# joint positions offsets
self.default_dof_pos = torch.zeros(
(self.num_envs, 12), dtype=torch.float, device=self.device, requires_grad=False
)
# reward episode sums
torch_zeros = lambda: torch.zeros(self.num_envs, dtype=torch.float, device=self.device, requires_grad=False)
self.episode_sums = {
"lin_vel_xy": torch_zeros(),
"lin_vel_z": torch_zeros(),
"ang_vel_z": torch_zeros(),
"ang_vel_xy": torch_zeros(),
"orient": torch_zeros(),
"torques": torch_zeros(),
"joint_acc": torch_zeros(),
"base_height": torch_zeros(),
"air_time": torch_zeros(),
"collision": torch_zeros(),
"stumble": torch_zeros(),
"action_rate": torch_zeros(),
"hip": torch_zeros(),
}
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
# normalization
self.lin_vel_scale = self._task_cfg["env"]["learn"]["linearVelocityScale"]
self.ang_vel_scale = self._task_cfg["env"]["learn"]["angularVelocityScale"]
self.dof_pos_scale = self._task_cfg["env"]["learn"]["dofPositionScale"]
self.dof_vel_scale = self._task_cfg["env"]["learn"]["dofVelocityScale"]
self.height_meas_scale = self._task_cfg["env"]["learn"]["heightMeasurementScale"]
self.action_scale = self._task_cfg["env"]["control"]["actionScale"]
# reward scales
self.rew_scales = {}
self.rew_scales["termination"] = self._task_cfg["env"]["learn"]["terminalReward"]
self.rew_scales["lin_vel_xy"] = self._task_cfg["env"]["learn"]["linearVelocityXYRewardScale"]
self.rew_scales["lin_vel_z"] = self._task_cfg["env"]["learn"]["linearVelocityZRewardScale"]
self.rew_scales["ang_vel_z"] = self._task_cfg["env"]["learn"]["angularVelocityZRewardScale"]
self.rew_scales["ang_vel_xy"] = self._task_cfg["env"]["learn"]["angularVelocityXYRewardScale"]
self.rew_scales["orient"] = self._task_cfg["env"]["learn"]["orientationRewardScale"]
self.rew_scales["torque"] = self._task_cfg["env"]["learn"]["torqueRewardScale"]
self.rew_scales["joint_acc"] = self._task_cfg["env"]["learn"]["jointAccRewardScale"]
self.rew_scales["base_height"] = self._task_cfg["env"]["learn"]["baseHeightRewardScale"]
self.rew_scales["action_rate"] = self._task_cfg["env"]["learn"]["actionRateRewardScale"]
self.rew_scales["hip"] = self._task_cfg["env"]["learn"]["hipRewardScale"]
self.rew_scales["fallen_over"] = self._task_cfg["env"]["learn"]["fallenOverRewardScale"]
# command ranges
self.command_x_range = self._task_cfg["env"]["randomCommandVelocityRanges"]["linear_x"]
self.command_y_range = self._task_cfg["env"]["randomCommandVelocityRanges"]["linear_y"]
self.command_yaw_range = self._task_cfg["env"]["randomCommandVelocityRanges"]["yaw"]
# base init state
pos = self._task_cfg["env"]["baseInitState"]["pos"]
rot = self._task_cfg["env"]["baseInitState"]["rot"]
v_lin = self._task_cfg["env"]["baseInitState"]["vLinear"]
v_ang = self._task_cfg["env"]["baseInitState"]["vAngular"]
self.base_init_state = pos + rot + v_lin + v_ang
# default joint positions
self.named_default_joint_angles = self._task_cfg["env"]["defaultJointAngles"]
# other
self.decimation = self._task_cfg["env"]["control"]["decimation"]
self.dt = self.decimation * self._task_cfg["sim"]["dt"]
self.max_episode_length_s = self._task_cfg["env"]["learn"]["episodeLength_s"]
self.max_episode_length = int(self.max_episode_length_s / self.dt + 0.5)
self.push_interval = int(self._task_cfg["env"]["learn"]["pushInterval_s"] / self.dt + 0.5)
self.Kp = self._task_cfg["env"]["control"]["stiffness"]
self.Kd = self._task_cfg["env"]["control"]["damping"]
self.curriculum = self._task_cfg["env"]["terrain"]["curriculum"]
self.base_threshold = 0.2
self.knee_threshold = 0.1
for key in self.rew_scales.keys():
self.rew_scales[key] *= self.dt
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._task_cfg["sim"]["default_physics_material"]["static_friction"] = self._task_cfg["env"]["terrain"][
"staticFriction"
]
self._task_cfg["sim"]["default_physics_material"]["dynamic_friction"] = self._task_cfg["env"]["terrain"][
"dynamicFriction"
]
self._task_cfg["sim"]["default_physics_material"]["restitution"] = self._task_cfg["env"]["terrain"][
"restitution"
]
self._task_cfg["sim"]["add_ground_plane"] = False
def _get_noise_scale_vec(self, cfg):
noise_vec = torch.zeros_like(self.obs_buf[0])
self.add_noise = self._task_cfg["env"]["learn"]["addNoise"]
noise_level = self._task_cfg["env"]["learn"]["noiseLevel"]
noise_vec[:3] = self._task_cfg["env"]["learn"]["linearVelocityNoise"] * noise_level * self.lin_vel_scale
noise_vec[3:6] = self._task_cfg["env"]["learn"]["angularVelocityNoise"] * noise_level * self.ang_vel_scale
noise_vec[6:9] = self._task_cfg["env"]["learn"]["gravityNoise"] * noise_level
noise_vec[9:12] = 0.0 # commands
noise_vec[12:24] = self._task_cfg["env"]["learn"]["dofPositionNoise"] * noise_level * self.dof_pos_scale
noise_vec[24:36] = self._task_cfg["env"]["learn"]["dofVelocityNoise"] * noise_level * self.dof_vel_scale
noise_vec[36:176] = (
self._task_cfg["env"]["learn"]["heightMeasurementNoise"] * noise_level * self.height_meas_scale
)
noise_vec[176:188] = 0.0 # previous actions
return noise_vec
def init_height_points(self):
# 1mx1.6m rectangle (without center line)
y = 0.1 * torch.tensor(
[-5, -4, -3, -2, -1, 1, 2, 3, 4, 5], device=self.device, requires_grad=False
) # 10-50cm on each side
x = 0.1 * torch.tensor(
[-8, -7, -6, -5, -4, -3, -2, 2, 3, 4, 5, 6, 7, 8], device=self.device, requires_grad=False
) # 20-80cm on each side
grid_x, grid_y = torch.meshgrid(x, y, indexing='ij')
self.num_height_points = grid_x.numel()
points = torch.zeros(self.num_envs, self.num_height_points, 3, device=self.device, requires_grad=False)
points[:, :, 0] = grid_x.flatten()
points[:, :, 1] = grid_y.flatten()
return points
def _create_trimesh(self, create_mesh=True):
self.terrain = Terrain(self._task_cfg["env"]["terrain"], num_robots=self.num_envs)
vertices = self.terrain.vertices
triangles = self.terrain.triangles
position = torch.tensor([-self.terrain.border_size, -self.terrain.border_size, 0.0])
if create_mesh:
add_terrain_to_stage(stage=self._stage, vertices=vertices, triangles=triangles, position=position)
self.height_samples = (
torch.tensor(self.terrain.heightsamples).view(self.terrain.tot_rows, self.terrain.tot_cols).to(self.device)
)
def set_up_scene(self, scene) -> None:
self._stage = get_current_stage()
self.get_terrain()
self.get_anymal()
super().set_up_scene(scene, collision_filter_global_paths=["/World/terrain"])
self._anymals = AnymalView(
prim_paths_expr="/World/envs/.*/anymal", name="anymal_view", track_contact_forces=True
)
scene.add(self._anymals)
scene.add(self._anymals._knees)
scene.add(self._anymals._base)
def initialize_views(self, scene):
# initialize terrain variables even if we do not need to re-create the terrain mesh
self.get_terrain(create_mesh=False)
super().initialize_views(scene)
if scene.object_exists("anymal_view"):
scene.remove_object("anymal_view", registry_only=True)
if scene.object_exists("knees_view"):
scene.remove_object("knees_view", registry_only=True)
if scene.object_exists("base_view"):
scene.remove_object("base_view", registry_only=True)
self._anymals = AnymalView(
prim_paths_expr="/World/envs/.*/anymal", name="anymal_view", track_contact_forces=True
)
scene.add(self._anymals)
scene.add(self._anymals._knees)
scene.add(self._anymals._base)
def get_terrain(self, create_mesh=True):
self.env_origins = torch.zeros((self.num_envs, 3), device=self.device, requires_grad=False)
if not self.curriculum:
self._task_cfg["env"]["terrain"]["maxInitMapLevel"] = self._task_cfg["env"]["terrain"]["numLevels"] - 1
self.terrain_levels = torch.randint(
0, self._task_cfg["env"]["terrain"]["maxInitMapLevel"] + 1, (self.num_envs,), device=self.device
)
self.terrain_types = torch.randint(
0, self._task_cfg["env"]["terrain"]["numTerrains"], (self.num_envs,), device=self.device
)
self._create_trimesh(create_mesh=create_mesh)
self.terrain_origins = torch.from_numpy(self.terrain.env_origins).to(self.device).to(torch.float)
def get_anymal(self):
anymal_translation = torch.tensor([0.0, 0.0, 0.66])
anymal_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0])
anymal = Anymal(
prim_path=self.default_zero_env_path + "/anymal",
name="anymal",
translation=anymal_translation,
orientation=anymal_orientation,
)
self._sim_config.apply_articulation_settings(
"anymal", get_prim_at_path(anymal.prim_path), self._sim_config.parse_actor_config("anymal")
)
anymal.set_anymal_properties(self._stage, anymal.prim)
anymal.prepare_contacts(self._stage, anymal.prim)
self.dof_names = anymal.dof_names
for i in range(self.num_actions):
name = self.dof_names[i]
angle = self.named_default_joint_angles[name]
self.default_dof_pos[:, i] = angle
def post_reset(self):
self.base_init_state = torch.tensor(
self.base_init_state, dtype=torch.float, device=self.device, requires_grad=False
)
self.timeout_buf = torch.zeros(self.num_envs, device=self.device, dtype=torch.long)
# initialize some data used later on
self.up_axis_idx = 2
self.common_step_counter = 0
self.extras = {}
self.noise_scale_vec = self._get_noise_scale_vec(self._task_cfg)
self.commands = torch.zeros(
self.num_envs, 4, dtype=torch.float, device=self.device, requires_grad=False
) # x vel, y vel, yaw vel, heading
self.commands_scale = torch.tensor(
[self.lin_vel_scale, self.lin_vel_scale, self.ang_vel_scale],
device=self.device,
requires_grad=False,
)
self.gravity_vec = torch.tensor(
get_axis_params(-1.0, self.up_axis_idx), dtype=torch.float, device=self.device
).repeat((self.num_envs, 1))
self.forward_vec = torch.tensor([1.0, 0.0, 0.0], dtype=torch.float, device=self.device).repeat(
(self.num_envs, 1)
)
self.torques = torch.zeros(
self.num_envs, self.num_actions, dtype=torch.float, device=self.device, requires_grad=False
)
self.actions = torch.zeros(
self.num_envs, self.num_actions, dtype=torch.float, device=self.device, requires_grad=False
)
self.last_actions = torch.zeros(
self.num_envs, self.num_actions, dtype=torch.float, device=self.device, requires_grad=False
)
self.feet_air_time = torch.zeros(self.num_envs, 4, dtype=torch.float, device=self.device, requires_grad=False)
self.last_dof_vel = torch.zeros((self.num_envs, 12), dtype=torch.float, device=self.device, requires_grad=False)
for i in range(self.num_envs):
self.env_origins[i] = self.terrain_origins[self.terrain_levels[i], self.terrain_types[i]]
self.num_dof = self._anymals.num_dof
self.dof_pos = torch.zeros((self.num_envs, self.num_dof), dtype=torch.float, device=self.device)
self.dof_vel = torch.zeros((self.num_envs, self.num_dof), dtype=torch.float, device=self.device)
self.base_pos = torch.zeros((self.num_envs, 3), dtype=torch.float, device=self.device)
self.base_quat = torch.zeros((self.num_envs, 4), dtype=torch.float, device=self.device)
self.base_velocities = torch.zeros((self.num_envs, 6), dtype=torch.float, device=self.device)
self.knee_pos = torch.zeros((self.num_envs * 4, 3), dtype=torch.float, device=self.device)
self.knee_quat = torch.zeros((self.num_envs * 4, 4), dtype=torch.float, device=self.device)
indices = torch.arange(self._num_envs, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
self.init_done = True
def reset_idx(self, env_ids):
indices = env_ids.to(dtype=torch.int32)
positions_offset = torch_rand_float(0.5, 1.5, (len(env_ids), self.num_dof), device=self.device)
velocities = torch_rand_float(-0.1, 0.1, (len(env_ids), self.num_dof), device=self.device)
self.dof_pos[env_ids] = self.default_dof_pos[env_ids] * positions_offset
self.dof_vel[env_ids] = velocities
self.update_terrain_level(env_ids)
self.base_pos[env_ids] = self.base_init_state[0:3]
self.base_pos[env_ids, 0:3] += self.env_origins[env_ids]
self.base_pos[env_ids, 0:2] += torch_rand_float(-0.5, 0.5, (len(env_ids), 2), device=self.device)
self.base_quat[env_ids] = self.base_init_state[3:7]
self.base_velocities[env_ids] = self.base_init_state[7:]
self._anymals.set_world_poses(
positions=self.base_pos[env_ids].clone(), orientations=self.base_quat[env_ids].clone(), indices=indices
)
self._anymals.set_velocities(velocities=self.base_velocities[env_ids].clone(), indices=indices)
self._anymals.set_joint_positions(positions=self.dof_pos[env_ids].clone(), indices=indices)
self._anymals.set_joint_velocities(velocities=self.dof_vel[env_ids].clone(), indices=indices)
self.commands[env_ids, 0] = torch_rand_float(
self.command_x_range[0], self.command_x_range[1], (len(env_ids), 1), device=self.device
).squeeze()
self.commands[env_ids, 1] = torch_rand_float(
self.command_y_range[0], self.command_y_range[1], (len(env_ids), 1), device=self.device
).squeeze()
self.commands[env_ids, 3] = torch_rand_float(
self.command_yaw_range[0], self.command_yaw_range[1], (len(env_ids), 1), device=self.device
).squeeze()
self.commands[env_ids] *= (torch.norm(self.commands[env_ids, :2], dim=1) > 0.25).unsqueeze(
1
) # set small commands to zero
self.last_actions[env_ids] = 0.0
self.last_dof_vel[env_ids] = 0.0
self.feet_air_time[env_ids] = 0.0
self.progress_buf[env_ids] = 0
self.reset_buf[env_ids] = 1
# fill extras
self.extras["episode"] = {}
for key in self.episode_sums.keys():
self.extras["episode"]["rew_" + key] = (
torch.mean(self.episode_sums[key][env_ids]) / self.max_episode_length_s
)
self.episode_sums[key][env_ids] = 0.0
self.extras["episode"]["terrain_level"] = torch.mean(self.terrain_levels.float())
def update_terrain_level(self, env_ids):
if not self.init_done or not self.curriculum:
# do not change on initial reset
return
root_pos, _ = self._anymals.get_world_poses(clone=False)
distance = torch.norm(root_pos[env_ids, :2] - self.env_origins[env_ids, :2], dim=1)
self.terrain_levels[env_ids] -= 1 * (
distance < torch.norm(self.commands[env_ids, :2]) * self.max_episode_length_s * 0.25
)
self.terrain_levels[env_ids] += 1 * (distance > self.terrain.env_length / 2)
self.terrain_levels[env_ids] = torch.clip(self.terrain_levels[env_ids], 0) % self.terrain.env_rows
self.env_origins[env_ids] = self.terrain_origins[self.terrain_levels[env_ids], self.terrain_types[env_ids]]
def refresh_dof_state_tensors(self):
self.dof_pos = self._anymals.get_joint_positions(clone=False)
self.dof_vel = self._anymals.get_joint_velocities(clone=False)
def refresh_body_state_tensors(self):
self.base_pos, self.base_quat = self._anymals.get_world_poses(clone=False)
self.base_velocities = self._anymals.get_velocities(clone=False)
self.knee_pos, self.knee_quat = self._anymals._knees.get_world_poses(clone=False)
def pre_physics_step(self, actions):
if not self._env._world.is_playing():
return
self.actions = actions.clone().to(self.device)
for i in range(self.decimation):
if self._env._world.is_playing():
torques = torch.clip(
self.Kp * (self.action_scale * self.actions + self.default_dof_pos - self.dof_pos)
- self.Kd * self.dof_vel,
-80.0,
80.0,
)
self._anymals.set_joint_efforts(torques)
self.torques = torques
SimulationContext.step(self._env._world, render=False)
self.refresh_dof_state_tensors()
def post_physics_step(self):
self.progress_buf[:] += 1
if self._env._world.is_playing():
self.refresh_dof_state_tensors()
self.refresh_body_state_tensors()
self.common_step_counter += 1
if self.common_step_counter % self.push_interval == 0:
self.push_robots()
# prepare quantities
self.base_lin_vel = quat_rotate_inverse(self.base_quat, self.base_velocities[:, 0:3])
self.base_ang_vel = quat_rotate_inverse(self.base_quat, self.base_velocities[:, 3:6])
self.projected_gravity = quat_rotate_inverse(self.base_quat, self.gravity_vec)
forward = quat_apply(self.base_quat, self.forward_vec)
heading = torch.atan2(forward[:, 1], forward[:, 0])
self.commands[:, 2] = torch.clip(0.5 * wrap_to_pi(self.commands[:, 3] - heading), -1.0, 1.0)
self.check_termination()
self.get_states()
self.calculate_metrics()
env_ids = self.reset_buf.nonzero(as_tuple=False).flatten()
if len(env_ids) > 0:
self.reset_idx(env_ids)
self.get_observations()
if self.add_noise:
self.obs_buf += (2 * torch.rand_like(self.obs_buf) - 1) * self.noise_scale_vec
self.last_actions[:] = self.actions[:]
self.last_dof_vel[:] = self.dof_vel[:]
return self.obs_buf, self.rew_buf, self.reset_buf, self.extras
def push_robots(self):
self.base_velocities[:, 0:2] = torch_rand_float(
-1.0, 1.0, (self.num_envs, 2), device=self.device
) # lin vel x/y
self._anymals.set_velocities(self.base_velocities)
def check_termination(self):
self.timeout_buf = torch.where(
self.progress_buf >= self.max_episode_length - 1,
torch.ones_like(self.timeout_buf),
torch.zeros_like(self.timeout_buf),
)
knee_contact = (
torch.norm(self._anymals._knees.get_net_contact_forces(clone=False).view(self._num_envs, 4, 3), dim=-1)
> 1.0
)
self.has_fallen = (torch.norm(self._anymals._base.get_net_contact_forces(clone=False), dim=1) > 1.0) | (
torch.sum(knee_contact, dim=-1) > 1.0
)
self.reset_buf = self.has_fallen.clone()
self.reset_buf = torch.where(self.timeout_buf.bool(), torch.ones_like(self.reset_buf), self.reset_buf)
def calculate_metrics(self):
# velocity tracking reward
lin_vel_error = torch.sum(torch.square(self.commands[:, :2] - self.base_lin_vel[:, :2]), dim=1)
ang_vel_error = torch.square(self.commands[:, 2] - self.base_ang_vel[:, 2])
rew_lin_vel_xy = torch.exp(-lin_vel_error / 0.25) * self.rew_scales["lin_vel_xy"]
rew_ang_vel_z = torch.exp(-ang_vel_error / 0.25) * self.rew_scales["ang_vel_z"]
# other base velocity penalties
rew_lin_vel_z = torch.square(self.base_lin_vel[:, 2]) * self.rew_scales["lin_vel_z"]
rew_ang_vel_xy = torch.sum(torch.square(self.base_ang_vel[:, :2]), dim=1) * self.rew_scales["ang_vel_xy"]
# orientation penalty
rew_orient = torch.sum(torch.square(self.projected_gravity[:, :2]), dim=1) * self.rew_scales["orient"]
# base height penalty
rew_base_height = torch.square(self.base_pos[:, 2] - 0.52) * self.rew_scales["base_height"]
# torque penalty
rew_torque = torch.sum(torch.square(self.torques), dim=1) * self.rew_scales["torque"]
# joint acc penalty
rew_joint_acc = torch.sum(torch.square(self.last_dof_vel - self.dof_vel), dim=1) * self.rew_scales["joint_acc"]
# fallen over penalty
rew_fallen_over = self.has_fallen * self.rew_scales["fallen_over"]
# action rate penalty
rew_action_rate = (
torch.sum(torch.square(self.last_actions - self.actions), dim=1) * self.rew_scales["action_rate"]
)
# cosmetic penalty for hip motion
rew_hip = (
torch.sum(torch.abs(self.dof_pos[:, 0:4] - self.default_dof_pos[:, 0:4]), dim=1) * self.rew_scales["hip"]
)
# total reward
self.rew_buf = (
rew_lin_vel_xy
+ rew_ang_vel_z
+ rew_lin_vel_z
+ rew_ang_vel_xy
+ rew_orient
+ rew_base_height
+ rew_torque
+ rew_joint_acc
+ rew_action_rate
+ rew_hip
+ rew_fallen_over
)
self.rew_buf = torch.clip(self.rew_buf, min=0.0, max=None)
# add termination reward
self.rew_buf += self.rew_scales["termination"] * self.reset_buf * ~self.timeout_buf
# log episode reward sums
self.episode_sums["lin_vel_xy"] += rew_lin_vel_xy
self.episode_sums["ang_vel_z"] += rew_ang_vel_z
self.episode_sums["lin_vel_z"] += rew_lin_vel_z
self.episode_sums["ang_vel_xy"] += rew_ang_vel_xy
self.episode_sums["orient"] += rew_orient
self.episode_sums["torques"] += rew_torque
self.episode_sums["joint_acc"] += rew_joint_acc
self.episode_sums["action_rate"] += rew_action_rate
self.episode_sums["base_height"] += rew_base_height
self.episode_sums["hip"] += rew_hip
def get_observations(self):
self.measured_heights = self.get_heights()
heights = (
torch.clip(self.base_pos[:, 2].unsqueeze(1) - 0.5 - self.measured_heights, -1, 1.0) * self.height_meas_scale
)
self.obs_buf = torch.cat(
(
self.base_lin_vel * self.lin_vel_scale,
self.base_ang_vel * self.ang_vel_scale,
self.projected_gravity,
self.commands[:, :3] * self.commands_scale,
self.dof_pos * self.dof_pos_scale,
self.dof_vel * self.dof_vel_scale,
heights,
self.actions,
),
dim=-1,
)
def get_ground_heights_below_knees(self):
points = self.knee_pos.reshape(self.num_envs, 4, 3)
points += self.terrain.border_size
points = (points / self.terrain.horizontal_scale).long()
px = points[:, :, 0].view(-1)
py = points[:, :, 1].view(-1)
px = torch.clip(px, 0, self.height_samples.shape[0] - 2)
py = torch.clip(py, 0, self.height_samples.shape[1] - 2)
heights1 = self.height_samples[px, py]
heights2 = self.height_samples[px + 1, py + 1]
heights = torch.min(heights1, heights2)
return heights.view(self.num_envs, -1) * self.terrain.vertical_scale
def get_ground_heights_below_base(self):
points = self.base_pos.reshape(self.num_envs, 1, 3)
points += self.terrain.border_size
points = (points / self.terrain.horizontal_scale).long()
px = points[:, :, 0].view(-1)
py = points[:, :, 1].view(-1)
px = torch.clip(px, 0, self.height_samples.shape[0] - 2)
py = torch.clip(py, 0, self.height_samples.shape[1] - 2)
heights1 = self.height_samples[px, py]
heights2 = self.height_samples[px + 1, py + 1]
heights = torch.min(heights1, heights2)
return heights.view(self.num_envs, -1) * self.terrain.vertical_scale
def get_heights(self, env_ids=None):
if env_ids:
points = quat_apply_yaw(
self.base_quat[env_ids].repeat(1, self.num_height_points), self.height_points[env_ids]
) + (self.base_pos[env_ids, 0:3]).unsqueeze(1)
else:
points = quat_apply_yaw(self.base_quat.repeat(1, self.num_height_points), self.height_points) + (
self.base_pos[:, 0:3]
).unsqueeze(1)
points += self.terrain.border_size
points = (points / self.terrain.horizontal_scale).long()
px = points[:, :, 0].view(-1)
py = points[:, :, 1].view(-1)
px = torch.clip(px, 0, self.height_samples.shape[0] - 2)
py = torch.clip(py, 0, self.height_samples.shape[1] - 2)
heights1 = self.height_samples[px, py]
heights2 = self.height_samples[px + 1, py + 1]
heights = torch.min(heights1, heights2)
return heights.view(self.num_envs, -1) * self.terrain.vertical_scale
@torch.jit.script
def quat_apply_yaw(quat, vec):
quat_yaw = quat.clone().view(-1, 4)
quat_yaw[:, 1:3] = 0.0
quat_yaw = normalize(quat_yaw)
return quat_apply(quat_yaw, vec)
@torch.jit.script
def wrap_to_pi(angles):
angles %= 2 * np.pi
angles -= 2 * np.pi * (angles > np.pi)
return angles
def get_axis_params(value, axis_idx, x_value=0.0, dtype=float, n_dims=3):
"""construct arguments to `Vec` according to axis index."""
zs = np.zeros((n_dims,))
assert axis_idx < n_dims, "the axis dim should be within the vector dimensions"
zs[axis_idx] = 1.0
params = np.where(zs == 1.0, value, zs)
params[0] = x_value
return list(params.astype(dtype))
| 29,337 | Python | 45.568254 | 120 | 0.609128 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/shadow_hand.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.shadow_hand import ShadowHand
from omniisaacgymenvs.robots.articulations.views.shadow_hand_view import ShadowHandView
from omniisaacgymenvs.tasks.shared.in_hand_manipulation import InHandManipulationTask
class ShadowHandTask(InHandManipulationTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
InHandManipulationTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self.object_type = self._task_cfg["env"]["objectType"]
assert self.object_type in ["block"]
self.obs_type = self._task_cfg["env"]["observationType"]
if not (self.obs_type in ["openai", "full_no_vel", "full", "full_state"]):
raise Exception(
"Unknown type of observations!\nobservationType should be one of: [openai, full_no_vel, full, full_state]"
)
print("Obs type:", self.obs_type)
self.num_obs_dict = {
"openai": 42,
"full_no_vel": 77,
"full": 157,
"full_state": 187,
}
self.asymmetric_obs = self._task_cfg["env"]["asymmetric_observations"]
self.use_vel_obs = False
self.fingertip_obs = True
self.fingertips = [
"robot0:ffdistal",
"robot0:mfdistal",
"robot0:rfdistal",
"robot0:lfdistal",
"robot0:thdistal",
]
self.num_fingertips = len(self.fingertips)
self.object_scale = torch.tensor([1.0, 1.0, 1.0])
self.force_torque_obs_scale = 10.0
num_states = 0
if self.asymmetric_obs:
num_states = 187
self._num_observations = self.num_obs_dict[self.obs_type]
self._num_actions = 20
self._num_states = num_states
InHandManipulationTask.update_config(self)
def get_starting_positions(self):
self.hand_start_translation = torch.tensor([0.0, 0.0, 0.5], device=self.device)
self.hand_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.pose_dy, self.pose_dz = -0.39, 0.10
def get_hand(self):
shadow_hand = ShadowHand(
prim_path=self.default_zero_env_path + "/shadow_hand",
name="shadow_hand",
translation=self.hand_start_translation,
orientation=self.hand_start_orientation,
)
self._sim_config.apply_articulation_settings(
"shadow_hand",
get_prim_at_path(shadow_hand.prim_path),
self._sim_config.parse_actor_config("shadow_hand"),
)
shadow_hand.set_shadow_hand_properties(stage=self._stage, shadow_hand_prim=shadow_hand.prim)
shadow_hand.set_motor_control_mode(stage=self._stage, shadow_hand_path=shadow_hand.prim_path)
def get_hand_view(self, scene):
hand_view = ShadowHandView(prim_paths_expr="/World/envs/.*/shadow_hand", name="shadow_hand_view")
scene.add(hand_view._fingers)
return hand_view
def get_observations(self):
self.get_object_goal_observations()
self.fingertip_pos, self.fingertip_rot = self._hands._fingers.get_world_poses(clone=False)
self.fingertip_pos -= self._env_pos.repeat((1, self.num_fingertips)).reshape(
self.num_envs * self.num_fingertips, 3
)
self.fingertip_velocities = self._hands._fingers.get_velocities(clone=False)
self.hand_dof_pos = self._hands.get_joint_positions(clone=False)
self.hand_dof_vel = self._hands.get_joint_velocities(clone=False)
if self.obs_type == "full_state" or self.asymmetric_obs:
self.vec_sensor_tensor = self._hands.get_measured_joint_forces(
joint_indices=self._hands._sensor_indices
).view(self._num_envs, -1)
if self.obs_type == "openai":
self.compute_fingertip_observations(True)
elif self.obs_type == "full_no_vel":
self.compute_full_observations(True)
elif self.obs_type == "full":
self.compute_full_observations()
elif self.obs_type == "full_state":
self.compute_full_state(False)
else:
print("Unkown observations type!")
if self.asymmetric_obs:
self.compute_full_state(True)
observations = {self._hands.name: {"obs_buf": self.obs_buf}}
return observations
def compute_fingertip_observations(self, no_vel=False):
if no_vel:
# Per https://arxiv.org/pdf/1808.00177.pdf Table 2
# Fingertip positions
# Object Position, but not orientation
# Relative target orientation
# 3*self.num_fingertips = 15
self.obs_buf[:, 0:15] = self.fingertip_pos.reshape(self.num_envs, 15)
self.obs_buf[:, 15:18] = self.object_pos
self.obs_buf[:, 18:22] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
self.obs_buf[:, 22:42] = self.actions
else:
# 13*self.num_fingertips = 65
self.obs_buf[:, 0:65] = self.fingertip_state.reshape(self.num_envs, 65)
self.obs_buf[:, 0:15] = self.fingertip_pos.reshape(self.num_envs, 3 * self.num_fingertips)
self.obs_buf[:, 15:35] = self.fingertip_rot.reshape(self.num_envs, 4 * self.num_fingertips)
self.obs_buf[:, 35:65] = self.fingertip_velocities.reshape(self.num_envs, 6 * self.num_fingertips)
self.obs_buf[:, 65:68] = self.object_pos
self.obs_buf[:, 68:72] = self.object_rot
self.obs_buf[:, 72:75] = self.object_linvel
self.obs_buf[:, 75:78] = self.vel_obs_scale * self.object_angvel
self.obs_buf[:, 78:81] = self.goal_pos
self.obs_buf[:, 81:85] = self.goal_rot
self.obs_buf[:, 85:89] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
self.obs_buf[:, 89:109] = self.actions
def compute_full_observations(self, no_vel=False):
if no_vel:
self.obs_buf[:, 0 : self.num_hand_dofs] = unscale(
self.hand_dof_pos, self.hand_dof_lower_limits, self.hand_dof_upper_limits
)
self.obs_buf[:, 24:37] = self.object_pos
self.obs_buf[:, 27:31] = self.object_rot
self.obs_buf[:, 31:34] = self.goal_pos
self.obs_buf[:, 34:38] = self.goal_rot
self.obs_buf[:, 38:42] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
self.obs_buf[:, 42:57] = self.fingertip_pos.reshape(self.num_envs, 3 * self.num_fingertips)
self.obs_buf[:, 57:77] = self.actions
else:
self.obs_buf[:, 0 : self.num_hand_dofs] = unscale(
self.hand_dof_pos, self.hand_dof_lower_limits, self.hand_dof_upper_limits
)
self.obs_buf[:, self.num_hand_dofs : 2 * self.num_hand_dofs] = self.vel_obs_scale * self.hand_dof_vel
self.obs_buf[:, 48:51] = self.object_pos
self.obs_buf[:, 51:55] = self.object_rot
self.obs_buf[:, 55:58] = self.object_linvel
self.obs_buf[:, 58:61] = self.vel_obs_scale * self.object_angvel
self.obs_buf[:, 61:64] = self.goal_pos
self.obs_buf[:, 64:68] = self.goal_rot
self.obs_buf[:, 68:72] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
# (7+6)*self.num_fingertips = 65
self.obs_buf[:, 72:87] = self.fingertip_pos.reshape(self.num_envs, 3 * self.num_fingertips)
self.obs_buf[:, 87:107] = self.fingertip_rot.reshape(self.num_envs, 4 * self.num_fingertips)
self.obs_buf[:, 107:137] = self.fingertip_velocities.reshape(self.num_envs, 6 * self.num_fingertips)
self.obs_buf[:, 137:157] = self.actions
def compute_full_state(self, asymm_obs=False):
if asymm_obs:
self.states_buf[:, 0 : self.num_hand_dofs] = unscale(
self.hand_dof_pos, self.hand_dof_lower_limits, self.hand_dof_upper_limits
)
self.states_buf[:, self.num_hand_dofs : 2 * self.num_hand_dofs] = self.vel_obs_scale * self.hand_dof_vel
# self.states_buf[:, 2*self.num_hand_dofs:3*self.num_hand_dofs] = self.force_torque_obs_scale * self.dof_force_tensor
obj_obs_start = 2 * self.num_hand_dofs # 48
self.states_buf[:, obj_obs_start : obj_obs_start + 3] = self.object_pos
self.states_buf[:, obj_obs_start + 3 : obj_obs_start + 7] = self.object_rot
self.states_buf[:, obj_obs_start + 7 : obj_obs_start + 10] = self.object_linvel
self.states_buf[:, obj_obs_start + 10 : obj_obs_start + 13] = self.vel_obs_scale * self.object_angvel
goal_obs_start = obj_obs_start + 13 # 61
self.states_buf[:, goal_obs_start : goal_obs_start + 3] = self.goal_pos
self.states_buf[:, goal_obs_start + 3 : goal_obs_start + 7] = self.goal_rot
self.states_buf[:, goal_obs_start + 7 : goal_obs_start + 11] = quat_mul(
self.object_rot, quat_conjugate(self.goal_rot)
)
# fingertip observations, state(pose and vel) + force-torque sensors
num_ft_states = 13 * self.num_fingertips # 65
num_ft_force_torques = 6 * self.num_fingertips # 30
fingertip_obs_start = goal_obs_start + 11 # 72
self.states_buf[
:, fingertip_obs_start : fingertip_obs_start + 3 * self.num_fingertips
] = self.fingertip_pos.reshape(self.num_envs, 3 * self.num_fingertips)
self.states_buf[
:, fingertip_obs_start + 3 * self.num_fingertips : fingertip_obs_start + 7 * self.num_fingertips
] = self.fingertip_rot.reshape(self.num_envs, 4 * self.num_fingertips)
self.states_buf[
:, fingertip_obs_start + 7 * self.num_fingertips : fingertip_obs_start + 13 * self.num_fingertips
] = self.fingertip_velocities.reshape(self.num_envs, 6 * self.num_fingertips)
self.states_buf[
:, fingertip_obs_start + num_ft_states : fingertip_obs_start + num_ft_states + num_ft_force_torques
] = (self.force_torque_obs_scale * self.vec_sensor_tensor)
# obs_end = 72 + 65 + 30 = 167
# obs_total = obs_end + num_actions = 187
obs_end = fingertip_obs_start + num_ft_states + num_ft_force_torques
self.states_buf[:, obs_end : obs_end + self.num_actions] = self.actions
else:
self.obs_buf[:, 0 : self.num_hand_dofs] = unscale(
self.hand_dof_pos, self.hand_dof_lower_limits, self.hand_dof_upper_limits
)
self.obs_buf[:, self.num_hand_dofs : 2 * self.num_hand_dofs] = self.vel_obs_scale * self.hand_dof_vel
self.obs_buf[:, 2 * self.num_hand_dofs : 3 * self.num_hand_dofs] = (
self.force_torque_obs_scale * self.dof_force_tensor
)
obj_obs_start = 3 * self.num_hand_dofs # 48
self.obs_buf[:, obj_obs_start : obj_obs_start + 3] = self.object_pos
self.obs_buf[:, obj_obs_start + 3 : obj_obs_start + 7] = self.object_rot
self.obs_buf[:, obj_obs_start + 7 : obj_obs_start + 10] = self.object_linvel
self.obs_buf[:, obj_obs_start + 10 : obj_obs_start + 13] = self.vel_obs_scale * self.object_angvel
goal_obs_start = obj_obs_start + 13 # 61
self.obs_buf[:, goal_obs_start : goal_obs_start + 3] = self.goal_pos
self.obs_buf[:, goal_obs_start + 3 : goal_obs_start + 7] = self.goal_rot
self.obs_buf[:, goal_obs_start + 7 : goal_obs_start + 11] = quat_mul(
self.object_rot, quat_conjugate(self.goal_rot)
)
# fingertip observations, state(pose and vel) + force-torque sensors
num_ft_states = 13 * self.num_fingertips # 65
num_ft_force_torques = 6 * self.num_fingertips # 30
fingertip_obs_start = goal_obs_start + 11 # 72
self.obs_buf[
:, fingertip_obs_start : fingertip_obs_start + 3 * self.num_fingertips
] = self.fingertip_pos.reshape(self.num_envs, 3 * self.num_fingertips)
self.obs_buf[
:, fingertip_obs_start + 3 * self.num_fingertips : fingertip_obs_start + 7 * self.num_fingertips
] = self.fingertip_rot.reshape(self.num_envs, 4 * self.num_fingertips)
self.obs_buf[
:, fingertip_obs_start + 7 * self.num_fingertips : fingertip_obs_start + 13 * self.num_fingertips
] = self.fingertip_velocities.reshape(self.num_envs, 6 * self.num_fingertips)
self.obs_buf[
:, fingertip_obs_start + num_ft_states : fingertip_obs_start + num_ft_states + num_ft_force_torques
] = (self.force_torque_obs_scale * self.vec_sensor_tensor)
# obs_end = 96 + 65 + 30 = 167
# obs_total = obs_end + num_actions = 187
obs_end = fingertip_obs_start + num_ft_states + num_ft_force_torques
self.obs_buf[:, obs_end : obs_end + self.num_actions] = self.actions
| 15,107 | Python | 48.211726 | 129 | 0.609188 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/franka_cabinet.py | # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
#
# NVIDIA CORPORATION and its licensors retain all intellectual property
# and proprietary rights in and to this software, related documentation
# and any modifications thereto. Any use, reproduction, disclosure or
# distribution of this software and related documentation without an express
# license agreement from NVIDIA CORPORATION is strictly prohibited.
#
import math
import numpy as np
import torch
from omni.isaac.cloner import Cloner
from omni.isaac.core.objects import DynamicCuboid
from omni.isaac.core.prims import RigidPrim, RigidPrimView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage
from omni.isaac.core.utils.torch.rotations import *
from omni.isaac.core.utils.torch.transformations import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.cabinet import Cabinet
from omniisaacgymenvs.robots.articulations.franka import Franka
from omniisaacgymenvs.robots.articulations.views.cabinet_view import CabinetView
from omniisaacgymenvs.robots.articulations.views.franka_view import FrankaView
from pxr import Usd, UsdGeom
class FrankaCabinetTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self.distX_offset = 0.04
self.dt = 1 / 60.0
self._num_observations = 23
self._num_actions = 9
RLTask.__init__(self, name, env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["episodeLength"]
self.action_scale = self._task_cfg["env"]["actionScale"]
self.start_position_noise = self._task_cfg["env"]["startPositionNoise"]
self.start_rotation_noise = self._task_cfg["env"]["startRotationNoise"]
self.num_props = self._task_cfg["env"]["numProps"]
self.dof_vel_scale = self._task_cfg["env"]["dofVelocityScale"]
self.dist_reward_scale = self._task_cfg["env"]["distRewardScale"]
self.rot_reward_scale = self._task_cfg["env"]["rotRewardScale"]
self.around_handle_reward_scale = self._task_cfg["env"]["aroundHandleRewardScale"]
self.open_reward_scale = self._task_cfg["env"]["openRewardScale"]
self.finger_dist_reward_scale = self._task_cfg["env"]["fingerDistRewardScale"]
self.action_penalty_scale = self._task_cfg["env"]["actionPenaltyScale"]
self.finger_close_reward_scale = self._task_cfg["env"]["fingerCloseRewardScale"]
def set_up_scene(self, scene) -> None:
self.get_franka()
self.get_cabinet()
if self.num_props > 0:
self.get_props()
super().set_up_scene(scene, filter_collisions=False)
self._frankas = FrankaView(prim_paths_expr="/World/envs/.*/franka", name="franka_view")
self._cabinets = CabinetView(prim_paths_expr="/World/envs/.*/cabinet", name="cabinet_view")
scene.add(self._frankas)
scene.add(self._frankas._hands)
scene.add(self._frankas._lfingers)
scene.add(self._frankas._rfingers)
scene.add(self._cabinets)
scene.add(self._cabinets._drawers)
if self.num_props > 0:
self._props = RigidPrimView(
prim_paths_expr="/World/envs/.*/prop/.*", name="prop_view", reset_xform_properties=False
)
scene.add(self._props)
self.init_data()
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("franka_view"):
scene.remove_object("franka_view", registry_only=True)
if scene.object_exists("hands_view"):
scene.remove_object("hands_view", registry_only=True)
if scene.object_exists("lfingers_view"):
scene.remove_object("lfingers_view", registry_only=True)
if scene.object_exists("rfingers_view"):
scene.remove_object("rfingers_view", registry_only=True)
if scene.object_exists("cabinet_view"):
scene.remove_object("cabinet_view", registry_only=True)
if scene.object_exists("drawers_view"):
scene.remove_object("drawers_view", registry_only=True)
if scene.object_exists("prop_view"):
scene.remove_object("prop_view", registry_only=True)
self._frankas = FrankaView(prim_paths_expr="/World/envs/.*/franka", name="franka_view")
self._cabinets = CabinetView(prim_paths_expr="/World/envs/.*/cabinet", name="cabinet_view")
scene.add(self._frankas)
scene.add(self._frankas._hands)
scene.add(self._frankas._lfingers)
scene.add(self._frankas._rfingers)
scene.add(self._cabinets)
scene.add(self._cabinets._drawers)
if self.num_props > 0:
self._props = RigidPrimView(
prim_paths_expr="/World/envs/.*/prop/.*", name="prop_view", reset_xform_properties=False
)
scene.add(self._props)
self.init_data()
def get_franka(self):
franka = Franka(prim_path=self.default_zero_env_path + "/franka", name="franka")
self._sim_config.apply_articulation_settings(
"franka", get_prim_at_path(franka.prim_path), self._sim_config.parse_actor_config("franka")
)
def get_cabinet(self):
cabinet = Cabinet(self.default_zero_env_path + "/cabinet", name="cabinet")
self._sim_config.apply_articulation_settings(
"cabinet", get_prim_at_path(cabinet.prim_path), self._sim_config.parse_actor_config("cabinet")
)
def get_props(self):
prop_cloner = Cloner()
drawer_pos = torch.tensor([0.0515, 0.0, 0.7172])
prop_color = torch.tensor([0.2, 0.4, 0.6])
props_per_row = int(math.ceil(math.sqrt(self.num_props)))
prop_size = 0.08
prop_spacing = 0.09
xmin = -0.5 * prop_spacing * (props_per_row - 1)
zmin = -0.5 * prop_spacing * (props_per_row - 1)
prop_count = 0
prop_pos = []
for j in range(props_per_row):
prop_up = zmin + j * prop_spacing
for k in range(props_per_row):
if prop_count >= self.num_props:
break
propx = xmin + k * prop_spacing
prop_pos.append([propx, prop_up, 0.0])
prop_count += 1
prop = DynamicCuboid(
prim_path=self.default_zero_env_path + "/prop/prop_0",
name="prop",
color=prop_color,
size=prop_size,
density=100.0,
)
self._sim_config.apply_articulation_settings(
"prop", get_prim_at_path(prop.prim_path), self._sim_config.parse_actor_config("prop")
)
prop_paths = [f"{self.default_zero_env_path}/prop/prop_{j}" for j in range(self.num_props)]
prop_cloner.clone(
source_prim_path=self.default_zero_env_path + "/prop/prop_0",
prim_paths=prop_paths,
positions=np.array(prop_pos) + drawer_pos.numpy(),
replicate_physics=False,
)
def init_data(self) -> None:
def get_env_local_pose(env_pos, xformable, device):
"""Compute pose in env-local coordinates"""
world_transform = xformable.ComputeLocalToWorldTransform(0)
world_pos = world_transform.ExtractTranslation()
world_quat = world_transform.ExtractRotationQuat()
px = world_pos[0] - env_pos[0]
py = world_pos[1] - env_pos[1]
pz = world_pos[2] - env_pos[2]
qx = world_quat.imaginary[0]
qy = world_quat.imaginary[1]
qz = world_quat.imaginary[2]
qw = world_quat.real
return torch.tensor([px, py, pz, qw, qx, qy, qz], device=device, dtype=torch.float)
stage = get_current_stage()
hand_pose = get_env_local_pose(
self._env_pos[0],
UsdGeom.Xformable(stage.GetPrimAtPath("/World/envs/env_0/franka/panda_link7")),
self._device,
)
lfinger_pose = get_env_local_pose(
self._env_pos[0],
UsdGeom.Xformable(stage.GetPrimAtPath("/World/envs/env_0/franka/panda_leftfinger")),
self._device,
)
rfinger_pose = get_env_local_pose(
self._env_pos[0],
UsdGeom.Xformable(stage.GetPrimAtPath("/World/envs/env_0/franka/panda_rightfinger")),
self._device,
)
finger_pose = torch.zeros(7, device=self._device)
finger_pose[0:3] = (lfinger_pose[0:3] + rfinger_pose[0:3]) / 2.0
finger_pose[3:7] = lfinger_pose[3:7]
hand_pose_inv_rot, hand_pose_inv_pos = tf_inverse(hand_pose[3:7], hand_pose[0:3])
grasp_pose_axis = 1
franka_local_grasp_pose_rot, franka_local_pose_pos = tf_combine(
hand_pose_inv_rot, hand_pose_inv_pos, finger_pose[3:7], finger_pose[0:3]
)
franka_local_pose_pos += torch.tensor([0, 0.04, 0], device=self._device)
self.franka_local_grasp_pos = franka_local_pose_pos.repeat((self._num_envs, 1))
self.franka_local_grasp_rot = franka_local_grasp_pose_rot.repeat((self._num_envs, 1))
drawer_local_grasp_pose = torch.tensor([0.3, 0.01, 0.0, 1.0, 0.0, 0.0, 0.0], device=self._device)
self.drawer_local_grasp_pos = drawer_local_grasp_pose[0:3].repeat((self._num_envs, 1))
self.drawer_local_grasp_rot = drawer_local_grasp_pose[3:7].repeat((self._num_envs, 1))
self.gripper_forward_axis = torch.tensor([0, 0, 1], device=self._device, dtype=torch.float).repeat(
(self._num_envs, 1)
)
self.drawer_inward_axis = torch.tensor([-1, 0, 0], device=self._device, dtype=torch.float).repeat(
(self._num_envs, 1)
)
self.gripper_up_axis = torch.tensor([0, 1, 0], device=self._device, dtype=torch.float).repeat(
(self._num_envs, 1)
)
self.drawer_up_axis = torch.tensor([0, 0, 1], device=self._device, dtype=torch.float).repeat(
(self._num_envs, 1)
)
self.franka_default_dof_pos = torch.tensor(
[1.157, -1.066, -0.155, -2.239, -1.841, 1.003, 0.469, 0.035, 0.035], device=self._device
)
self.actions = torch.zeros((self._num_envs, self.num_actions), device=self._device)
def get_observations(self) -> dict:
hand_pos, hand_rot = self._frankas._hands.get_world_poses(clone=False)
drawer_pos, drawer_rot = self._cabinets._drawers.get_world_poses(clone=False)
franka_dof_pos = self._frankas.get_joint_positions(clone=False)
franka_dof_vel = self._frankas.get_joint_velocities(clone=False)
self.cabinet_dof_pos = self._cabinets.get_joint_positions(clone=False)
self.cabinet_dof_vel = self._cabinets.get_joint_velocities(clone=False)
self.franka_dof_pos = franka_dof_pos
(
self.franka_grasp_rot,
self.franka_grasp_pos,
self.drawer_grasp_rot,
self.drawer_grasp_pos,
) = self.compute_grasp_transforms(
hand_rot,
hand_pos,
self.franka_local_grasp_rot,
self.franka_local_grasp_pos,
drawer_rot,
drawer_pos,
self.drawer_local_grasp_rot,
self.drawer_local_grasp_pos,
)
self.franka_lfinger_pos, self.franka_lfinger_rot = self._frankas._lfingers.get_world_poses(clone=False)
self.franka_rfinger_pos, self.franka_rfinger_rot = self._frankas._lfingers.get_world_poses(clone=False)
dof_pos_scaled = (
2.0
* (franka_dof_pos - self.franka_dof_lower_limits)
/ (self.franka_dof_upper_limits - self.franka_dof_lower_limits)
- 1.0
)
to_target = self.drawer_grasp_pos - self.franka_grasp_pos
self.obs_buf = torch.cat(
(
dof_pos_scaled,
franka_dof_vel * self.dof_vel_scale,
to_target,
self.cabinet_dof_pos[:, 3].unsqueeze(-1),
self.cabinet_dof_vel[:, 3].unsqueeze(-1),
),
dim=-1,
)
observations = {self._frankas.name: {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
self.actions = actions.clone().to(self._device)
targets = self.franka_dof_targets + self.franka_dof_speed_scales * self.dt * self.actions * self.action_scale
self.franka_dof_targets[:] = tensor_clamp(targets, self.franka_dof_lower_limits, self.franka_dof_upper_limits)
env_ids_int32 = torch.arange(self._frankas.count, dtype=torch.int32, device=self._device)
self._frankas.set_joint_position_targets(self.franka_dof_targets, indices=env_ids_int32)
def reset_idx(self, env_ids):
indices = env_ids.to(dtype=torch.int32)
num_indices = len(indices)
# reset franka
pos = tensor_clamp(
self.franka_default_dof_pos.unsqueeze(0)
+ 0.25 * (torch.rand((len(env_ids), self.num_franka_dofs), device=self._device) - 0.5),
self.franka_dof_lower_limits,
self.franka_dof_upper_limits,
)
dof_pos = torch.zeros((num_indices, self._frankas.num_dof), device=self._device)
dof_vel = torch.zeros((num_indices, self._frankas.num_dof), device=self._device)
dof_pos[:, :] = pos
self.franka_dof_targets[env_ids, :] = pos
self.franka_dof_pos[env_ids, :] = pos
# reset cabinet
self._cabinets.set_joint_positions(
torch.zeros_like(self._cabinets.get_joint_positions(clone=False)[env_ids]), indices=indices
)
self._cabinets.set_joint_velocities(
torch.zeros_like(self._cabinets.get_joint_velocities(clone=False)[env_ids]), indices=indices
)
# reset props
if self.num_props > 0:
self._props.set_world_poses(
self.default_prop_pos[self.prop_indices[env_ids].flatten()],
self.default_prop_rot[self.prop_indices[env_ids].flatten()],
self.prop_indices[env_ids].flatten().to(torch.int32),
)
self._frankas.set_joint_position_targets(self.franka_dof_targets[env_ids], indices=indices)
self._frankas.set_joint_positions(dof_pos, indices=indices)
self._frankas.set_joint_velocities(dof_vel, indices=indices)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def post_reset(self):
self.num_franka_dofs = self._frankas.num_dof
self.franka_dof_pos = torch.zeros((self.num_envs, self.num_franka_dofs), device=self._device)
dof_limits = self._frankas.get_dof_limits()
self.franka_dof_lower_limits = dof_limits[0, :, 0].to(device=self._device)
self.franka_dof_upper_limits = dof_limits[0, :, 1].to(device=self._device)
self.franka_dof_speed_scales = torch.ones_like(self.franka_dof_lower_limits)
self.franka_dof_speed_scales[self._frankas.gripper_indices] = 0.1
self.franka_dof_targets = torch.zeros(
(self._num_envs, self.num_franka_dofs), dtype=torch.float, device=self._device
)
if self.num_props > 0:
self.default_prop_pos, self.default_prop_rot = self._props.get_world_poses()
self.prop_indices = torch.arange(self._num_envs * self.num_props, device=self._device).view(
self._num_envs, self.num_props
)
# randomize all envs
indices = torch.arange(self._num_envs, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
def calculate_metrics(self) -> None:
self.rew_buf[:] = self.compute_franka_reward(
self.reset_buf,
self.progress_buf,
self.actions,
self.cabinet_dof_pos,
self.franka_grasp_pos,
self.drawer_grasp_pos,
self.franka_grasp_rot,
self.drawer_grasp_rot,
self.franka_lfinger_pos,
self.franka_rfinger_pos,
self.gripper_forward_axis,
self.drawer_inward_axis,
self.gripper_up_axis,
self.drawer_up_axis,
self._num_envs,
self.dist_reward_scale,
self.rot_reward_scale,
self.around_handle_reward_scale,
self.open_reward_scale,
self.finger_dist_reward_scale,
self.action_penalty_scale,
self.distX_offset,
self._max_episode_length,
self.franka_dof_pos,
self.finger_close_reward_scale,
)
def is_done(self) -> None:
# reset if drawer is open or max length reached
self.reset_buf = torch.where(self.cabinet_dof_pos[:, 3] > 0.39, torch.ones_like(self.reset_buf), self.reset_buf)
self.reset_buf = torch.where(
self.progress_buf >= self._max_episode_length - 1, torch.ones_like(self.reset_buf), self.reset_buf
)
def compute_grasp_transforms(
self,
hand_rot,
hand_pos,
franka_local_grasp_rot,
franka_local_grasp_pos,
drawer_rot,
drawer_pos,
drawer_local_grasp_rot,
drawer_local_grasp_pos,
):
global_franka_rot, global_franka_pos = tf_combine(
hand_rot, hand_pos, franka_local_grasp_rot, franka_local_grasp_pos
)
global_drawer_rot, global_drawer_pos = tf_combine(
drawer_rot, drawer_pos, drawer_local_grasp_rot, drawer_local_grasp_pos
)
return global_franka_rot, global_franka_pos, global_drawer_rot, global_drawer_pos
def compute_franka_reward(
self,
reset_buf,
progress_buf,
actions,
cabinet_dof_pos,
franka_grasp_pos,
drawer_grasp_pos,
franka_grasp_rot,
drawer_grasp_rot,
franka_lfinger_pos,
franka_rfinger_pos,
gripper_forward_axis,
drawer_inward_axis,
gripper_up_axis,
drawer_up_axis,
num_envs,
dist_reward_scale,
rot_reward_scale,
around_handle_reward_scale,
open_reward_scale,
finger_dist_reward_scale,
action_penalty_scale,
distX_offset,
max_episode_length,
joint_positions,
finger_close_reward_scale,
):
# type: (Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, int, float, float, float, float, float, float, float, float, Tensor) -> Tuple[Tensor, Tensor]
# distance from hand to the drawer
d = torch.norm(franka_grasp_pos - drawer_grasp_pos, p=2, dim=-1)
dist_reward = 1.0 / (1.0 + d**2)
dist_reward *= dist_reward
dist_reward = torch.where(d <= 0.02, dist_reward * 2, dist_reward)
axis1 = tf_vector(franka_grasp_rot, gripper_forward_axis)
axis2 = tf_vector(drawer_grasp_rot, drawer_inward_axis)
axis3 = tf_vector(franka_grasp_rot, gripper_up_axis)
axis4 = tf_vector(drawer_grasp_rot, drawer_up_axis)
dot1 = (
torch.bmm(axis1.view(num_envs, 1, 3), axis2.view(num_envs, 3, 1)).squeeze(-1).squeeze(-1)
) # alignment of forward axis for gripper
dot2 = (
torch.bmm(axis3.view(num_envs, 1, 3), axis4.view(num_envs, 3, 1)).squeeze(-1).squeeze(-1)
) # alignment of up axis for gripper
# reward for matching the orientation of the hand to the drawer (fingers wrapped)
rot_reward = 0.5 * (torch.sign(dot1) * dot1**2 + torch.sign(dot2) * dot2**2)
# bonus if left finger is above the drawer handle and right below
around_handle_reward = torch.zeros_like(rot_reward)
around_handle_reward = torch.where(
franka_lfinger_pos[:, 2] > drawer_grasp_pos[:, 2],
torch.where(
franka_rfinger_pos[:, 2] < drawer_grasp_pos[:, 2], around_handle_reward + 0.5, around_handle_reward
),
around_handle_reward,
)
# reward for distance of each finger from the drawer
finger_dist_reward = torch.zeros_like(rot_reward)
lfinger_dist = torch.abs(franka_lfinger_pos[:, 2] - drawer_grasp_pos[:, 2])
rfinger_dist = torch.abs(franka_rfinger_pos[:, 2] - drawer_grasp_pos[:, 2])
finger_dist_reward = torch.where(
franka_lfinger_pos[:, 2] > drawer_grasp_pos[:, 2],
torch.where(
franka_rfinger_pos[:, 2] < drawer_grasp_pos[:, 2],
(0.04 - lfinger_dist) + (0.04 - rfinger_dist),
finger_dist_reward,
),
finger_dist_reward,
)
finger_close_reward = torch.zeros_like(rot_reward)
finger_close_reward = torch.where(
d <= 0.03, (0.04 - joint_positions[:, 7]) + (0.04 - joint_positions[:, 8]), finger_close_reward
)
# regularization on the actions (summed for each environment)
action_penalty = torch.sum(actions**2, dim=-1)
# how far the cabinet has been opened out
open_reward = cabinet_dof_pos[:, 3] * around_handle_reward + cabinet_dof_pos[:, 3] # drawer_top_joint
rewards = (
dist_reward_scale * dist_reward
+ rot_reward_scale * rot_reward
+ around_handle_reward_scale * around_handle_reward
+ open_reward_scale * open_reward
+ finger_dist_reward_scale * finger_dist_reward
- action_penalty_scale * action_penalty
+ finger_close_reward * finger_close_reward_scale
)
# bonus for opening drawer properly
rewards = torch.where(cabinet_dof_pos[:, 3] > 0.01, rewards + 0.5, rewards)
rewards = torch.where(cabinet_dof_pos[:, 3] > 0.2, rewards + around_handle_reward, rewards)
rewards = torch.where(cabinet_dof_pos[:, 3] > 0.39, rewards + (2.0 * around_handle_reward), rewards)
# # prevent bad style in opening drawer
# rewards = torch.where(franka_lfinger_pos[:, 0] < drawer_grasp_pos[:, 0] - distX_offset,
# torch.ones_like(rewards) * -1, rewards)
# rewards = torch.where(franka_rfinger_pos[:, 0] < drawer_grasp_pos[:, 0] - distX_offset,
# torch.ones_like(rewards) * -1, rewards)
return rewards
| 22,939 | Python | 41.324723 | 222 | 0.599895 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/crazyflie.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import numpy as np
import torch
from omni.isaac.core.objects import DynamicSphere
from omni.isaac.core.prims import RigidPrimView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch.rotations import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.crazyflie import Crazyflie
from omniisaacgymenvs.robots.articulations.views.crazyflie_view import CrazyflieView
EPS = 1e-6 # small constant to avoid divisions by 0 and log(0)
class CrazyflieTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._num_observations = 18
self._num_actions = 4
self._crazyflie_position = torch.tensor([0, 0, 1.0])
self._ball_position = torch.tensor([0, 0, 1.0])
RLTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["maxEpisodeLength"]
self.dt = self._task_cfg["sim"]["dt"]
# parameters for the crazyflie
self.arm_length = 0.05
# parameters for the controller
self.motor_damp_time_up = 0.15
self.motor_damp_time_down = 0.15
# I use the multiplier 4, since 4*T ~ time for a step response to finish, where
# T is a time constant of the first-order filter
self.motor_tau_up = 4 * self.dt / (self.motor_damp_time_up + EPS)
self.motor_tau_down = 4 * self.dt / (self.motor_damp_time_down + EPS)
# thrust max
self.mass = 0.028
self.thrust_to_weight = 1.9
self.motor_assymetry = np.array([1.0, 1.0, 1.0, 1.0])
# re-normalizing to sum-up to 4
self.motor_assymetry = self.motor_assymetry * 4.0 / np.sum(self.motor_assymetry)
self.grav_z = -1.0 * self._task_cfg["sim"]["gravity"][2]
def set_up_scene(self, scene) -> None:
self.get_crazyflie()
self.get_target()
RLTask.set_up_scene(self, scene)
self._copters = CrazyflieView(prim_paths_expr="/World/envs/.*/Crazyflie", name="crazyflie_view")
self._balls = RigidPrimView(prim_paths_expr="/World/envs/.*/ball", name="ball_view")
scene.add(self._copters)
scene.add(self._balls)
for i in range(4):
scene.add(self._copters.physics_rotors[i])
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("crazyflie_view"):
scene.remove_object("crazyflie_view", registry_only=True)
if scene.object_exists("ball_view"):
scene.remove_object("ball_view", registry_only=True)
for i in range(1, 5):
scene.remove_object(f"m{i}_prop_view", registry_only=True)
self._copters = CrazyflieView(prim_paths_expr="/World/envs/.*/Crazyflie", name="crazyflie_view")
self._balls = RigidPrimView(prim_paths_expr="/World/envs/.*/ball", name="ball_view")
scene.add(self._copters)
scene.add(self._balls)
for i in range(4):
scene.add(self._copters.physics_rotors[i])
def get_crazyflie(self):
copter = Crazyflie(
prim_path=self.default_zero_env_path + "/Crazyflie", name="crazyflie", translation=self._crazyflie_position
)
self._sim_config.apply_articulation_settings(
"crazyflie", get_prim_at_path(copter.prim_path), self._sim_config.parse_actor_config("crazyflie")
)
def get_target(self):
radius = 0.2
color = torch.tensor([1, 0, 0])
ball = DynamicSphere(
prim_path=self.default_zero_env_path + "/ball",
translation=self._ball_position,
name="target_0",
radius=radius,
color=color,
)
self._sim_config.apply_articulation_settings(
"ball", get_prim_at_path(ball.prim_path), self._sim_config.parse_actor_config("ball")
)
ball.set_collision_enabled(False)
def get_observations(self) -> dict:
self.root_pos, self.root_rot = self._copters.get_world_poses(clone=False)
self.root_velocities = self._copters.get_velocities(clone=False)
root_positions = self.root_pos - self._env_pos
root_quats = self.root_rot
rot_x = quat_axis(root_quats, 0)
rot_y = quat_axis(root_quats, 1)
rot_z = quat_axis(root_quats, 2)
root_linvels = self.root_velocities[:, :3]
root_angvels = self.root_velocities[:, 3:]
self.obs_buf[..., 0:3] = self.target_positions - root_positions
self.obs_buf[..., 3:6] = rot_x
self.obs_buf[..., 6:9] = rot_y
self.obs_buf[..., 9:12] = rot_z
self.obs_buf[..., 12:15] = root_linvels
self.obs_buf[..., 15:18] = root_angvels
observations = {self._copters.name: {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
set_target_ids = (self.progress_buf % 500 == 0).nonzero(as_tuple=False).squeeze(-1)
if len(set_target_ids) > 0:
self.set_targets(set_target_ids)
actions = actions.clone().to(self._device)
self.actions = actions
# clamp to [-1.0, 1.0]
thrust_cmds = torch.clamp(actions, min=-1.0, max=1.0)
# scale to [0.0, 1.0]
thrust_cmds = (thrust_cmds + 1.0) / 2.0
# filtering the thruster and adding noise
motor_tau = self.motor_tau_up * torch.ones((self._num_envs, 4), dtype=torch.float32, device=self._device)
motor_tau[thrust_cmds < self.thrust_cmds_damp] = self.motor_tau_down
motor_tau[motor_tau > 1.0] = 1.0
# Since NN commands thrusts we need to convert to rot vel and back
thrust_rot = thrust_cmds**0.5
self.thrust_rot_damp = motor_tau * (thrust_rot - self.thrust_rot_damp) + self.thrust_rot_damp
self.thrust_cmds_damp = self.thrust_rot_damp**2
## Adding noise
thrust_noise = 0.01 * torch.randn(4, dtype=torch.float32, device=self._device)
thrust_noise = thrust_cmds * thrust_noise
self.thrust_cmds_damp = torch.clamp(self.thrust_cmds_damp + thrust_noise, min=0.0, max=1.0)
thrusts = self.thrust_max * self.thrust_cmds_damp
# thrusts given rotation
root_quats = self.root_rot
rot_x = quat_axis(root_quats, 0)
rot_y = quat_axis(root_quats, 1)
rot_z = quat_axis(root_quats, 2)
rot_matrix = torch.cat((rot_x, rot_y, rot_z), 1).reshape(-1, 3, 3)
force_x = torch.zeros(self._num_envs, 4, dtype=torch.float32, device=self._device)
force_y = torch.zeros(self._num_envs, 4, dtype=torch.float32, device=self._device)
force_xy = torch.cat((force_x, force_y), 1).reshape(-1, 4, 2)
thrusts = thrusts.reshape(-1, 4, 1)
thrusts = torch.cat((force_xy, thrusts), 2)
thrusts_0 = thrusts[:, 0]
thrusts_0 = thrusts_0[:, :, None]
thrusts_1 = thrusts[:, 1]
thrusts_1 = thrusts_1[:, :, None]
thrusts_2 = thrusts[:, 2]
thrusts_2 = thrusts_2[:, :, None]
thrusts_3 = thrusts[:, 3]
thrusts_3 = thrusts_3[:, :, None]
mod_thrusts_0 = torch.matmul(rot_matrix, thrusts_0)
mod_thrusts_1 = torch.matmul(rot_matrix, thrusts_1)
mod_thrusts_2 = torch.matmul(rot_matrix, thrusts_2)
mod_thrusts_3 = torch.matmul(rot_matrix, thrusts_3)
self.thrusts[:, 0] = torch.squeeze(mod_thrusts_0)
self.thrusts[:, 1] = torch.squeeze(mod_thrusts_1)
self.thrusts[:, 2] = torch.squeeze(mod_thrusts_2)
self.thrusts[:, 3] = torch.squeeze(mod_thrusts_3)
# clear actions for reset envs
self.thrusts[reset_env_ids] = 0
# spin spinning rotors
prop_rot = self.thrust_cmds_damp * self.prop_max_rot
self.dof_vel[:, 0] = prop_rot[:, 0]
self.dof_vel[:, 1] = -1.0 * prop_rot[:, 1]
self.dof_vel[:, 2] = prop_rot[:, 2]
self.dof_vel[:, 3] = -1.0 * prop_rot[:, 3]
self._copters.set_joint_velocities(self.dof_vel)
# apply actions
for i in range(4):
self._copters.physics_rotors[i].apply_forces(self.thrusts[:, i], indices=self.all_indices)
def post_reset(self):
thrust_max = self.grav_z * self.mass * self.thrust_to_weight * self.motor_assymetry / 4.0
self.thrusts = torch.zeros((self._num_envs, 4, 3), dtype=torch.float32, device=self._device)
self.thrust_cmds_damp = torch.zeros((self._num_envs, 4), dtype=torch.float32, device=self._device)
self.thrust_rot_damp = torch.zeros((self._num_envs, 4), dtype=torch.float32, device=self._device)
self.thrust_max = torch.tensor(thrust_max, device=self._device, dtype=torch.float32)
self.motor_linearity = 1.0
self.prop_max_rot = 433.3
self.target_positions = torch.zeros((self._num_envs, 3), device=self._device, dtype=torch.float32)
self.target_positions[:, 2] = 1
self.actions = torch.zeros((self._num_envs, 4), device=self._device, dtype=torch.float32)
self.all_indices = torch.arange(self._num_envs, dtype=torch.int32, device=self._device)
# Extra info
self.extras = {}
torch_zeros = lambda: torch.zeros(self.num_envs, dtype=torch.float, device=self.device, requires_grad=False)
self.episode_sums = {
"rew_pos": torch_zeros(),
"rew_orient": torch_zeros(),
"rew_effort": torch_zeros(),
"rew_spin": torch_zeros(),
"raw_dist": torch_zeros(),
"raw_orient": torch_zeros(),
"raw_effort": torch_zeros(),
"raw_spin": torch_zeros(),
}
self.root_pos, self.root_rot = self._copters.get_world_poses()
self.root_velocities = self._copters.get_velocities()
self.dof_pos = self._copters.get_joint_positions()
self.dof_vel = self._copters.get_joint_velocities()
self.initial_ball_pos, self.initial_ball_rot = self._balls.get_world_poses(clone=False)
self.initial_root_pos, self.initial_root_rot = self.root_pos.clone(), self.root_rot.clone()
# control parameters
self.thrusts = torch.zeros((self._num_envs, 4, 3), dtype=torch.float32, device=self._device)
self.thrust_cmds_damp = torch.zeros((self._num_envs, 4), dtype=torch.float32, device=self._device)
self.thrust_rot_damp = torch.zeros((self._num_envs, 4), dtype=torch.float32, device=self._device)
self.set_targets(self.all_indices)
def set_targets(self, env_ids):
num_sets = len(env_ids)
envs_long = env_ids.long()
# set target position randomly with x, y in (0, 0) and z in (2)
self.target_positions[envs_long, 0:2] = torch.zeros((num_sets, 2), device=self._device)
self.target_positions[envs_long, 2] = torch.ones(num_sets, device=self._device) * 2.0
# shift the target up so it visually aligns better
ball_pos = self.target_positions[envs_long] + self._env_pos[envs_long]
ball_pos[:, 2] += 0.0
self._balls.set_world_poses(ball_pos[:, 0:3], self.initial_ball_rot[envs_long].clone(), indices=env_ids)
def reset_idx(self, env_ids):
num_resets = len(env_ids)
self.dof_pos[env_ids, :] = torch_rand_float(-0.0, 0.0, (num_resets, self._copters.num_dof), device=self._device)
self.dof_vel[env_ids, :] = 0
root_pos = self.initial_root_pos.clone()
root_pos[env_ids, 0] += torch_rand_float(-0.0, 0.0, (num_resets, 1), device=self._device).view(-1)
root_pos[env_ids, 1] += torch_rand_float(-0.0, 0.0, (num_resets, 1), device=self._device).view(-1)
root_pos[env_ids, 2] += torch_rand_float(-0.0, 0.0, (num_resets, 1), device=self._device).view(-1)
root_velocities = self.root_velocities.clone()
root_velocities[env_ids] = 0
# apply resets
self._copters.set_joint_positions(self.dof_pos[env_ids], indices=env_ids)
self._copters.set_joint_velocities(self.dof_vel[env_ids], indices=env_ids)
self._copters.set_world_poses(root_pos[env_ids], self.initial_root_rot[env_ids].clone(), indices=env_ids)
self._copters.set_velocities(root_velocities[env_ids], indices=env_ids)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
self.thrust_cmds_damp[env_ids] = 0
self.thrust_rot_damp[env_ids] = 0
# fill extras
self.extras["episode"] = {}
for key in self.episode_sums.keys():
self.extras["episode"][key] = torch.mean(self.episode_sums[key][env_ids]) / self._max_episode_length
self.episode_sums[key][env_ids] = 0.0
def calculate_metrics(self) -> None:
root_positions = self.root_pos - self._env_pos
root_quats = self.root_rot
root_angvels = self.root_velocities[:, 3:]
# pos reward
target_dist = torch.sqrt(torch.square(self.target_positions - root_positions).sum(-1))
pos_reward = 1.0 / (1.0 + target_dist)
self.target_dist = target_dist
self.root_positions = root_positions
# orient reward
ups = quat_axis(root_quats, 2)
self.orient_z = ups[..., 2]
up_reward = torch.clamp(ups[..., 2], min=0.0, max=1.0)
# effort reward
effort = torch.square(self.actions).sum(-1)
effort_reward = 0.05 * torch.exp(-0.5 * effort)
# spin reward
spin = torch.square(root_angvels).sum(-1)
spin_reward = 0.01 * torch.exp(-1.0 * spin)
# combined reward
self.rew_buf[:] = pos_reward + pos_reward * (up_reward + spin_reward) - effort_reward
# log episode reward sums
self.episode_sums["rew_pos"] += pos_reward
self.episode_sums["rew_orient"] += up_reward
self.episode_sums["rew_effort"] += effort_reward
self.episode_sums["rew_spin"] += spin_reward
# log raw info
self.episode_sums["raw_dist"] += target_dist
self.episode_sums["raw_orient"] += ups[..., 2]
self.episode_sums["raw_effort"] += effort
self.episode_sums["raw_spin"] += spin
def is_done(self) -> None:
# resets due to misbehavior
ones = torch.ones_like(self.reset_buf)
die = torch.zeros_like(self.reset_buf)
die = torch.where(self.target_dist > 5.0, ones, die)
# z >= 0.5 & z <= 5.0 & up > 0
die = torch.where(self.root_positions[..., 2] < 0.5, ones, die)
die = torch.where(self.root_positions[..., 2] > 5.0, ones, die)
die = torch.where(self.orient_z < 0.0, ones, die)
# resets due to episode length
self.reset_buf[:] = torch.where(self.progress_buf >= self._max_episode_length - 1, ones, die)
| 16,830 | Python | 41.502525 | 120 | 0.61937 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/humanoid.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch.maths import tensor_clamp, torch_rand_float, unscale
from omni.isaac.core.utils.torch.rotations import compute_heading_and_up, compute_rot, quat_conjugate
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.humanoid import Humanoid
from omniisaacgymenvs.tasks.shared.locomotion import LocomotionTask
from pxr import PhysxSchema
class HumanoidLocomotionTask(LocomotionTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._num_observations = 87
self._num_actions = 21
self._humanoid_positions = torch.tensor([0, 0, 1.34])
LocomotionTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
LocomotionTask.update_config(self)
def set_up_scene(self, scene) -> None:
self.get_humanoid()
RLTask.set_up_scene(self, scene)
self._humanoids = ArticulationView(
prim_paths_expr="/World/envs/.*/Humanoid/torso", name="humanoid_view", reset_xform_properties=False
)
scene.add(self._humanoids)
return
def initialize_views(self, scene):
RLTask.initialize_views(self, scene)
if scene.object_exists("humanoid_view"):
scene.remove_object("humanoid_view", registry_only=True)
self._humanoids = ArticulationView(
prim_paths_expr="/World/envs/.*/Humanoid/torso", name="humanoid_view", reset_xform_properties=False
)
scene.add(self._humanoids)
def get_humanoid(self):
humanoid = Humanoid(
prim_path=self.default_zero_env_path + "/Humanoid", name="Humanoid", translation=self._humanoid_positions
)
self._sim_config.apply_articulation_settings(
"Humanoid", get_prim_at_path(humanoid.prim_path), self._sim_config.parse_actor_config("Humanoid")
)
def get_robot(self):
return self._humanoids
def post_reset(self):
self.joint_gears = torch.tensor(
[
67.5000, # lower_waist
67.5000, # lower_waist
67.5000, # right_upper_arm
67.5000, # right_upper_arm
67.5000, # left_upper_arm
67.5000, # left_upper_arm
67.5000, # pelvis
45.0000, # right_lower_arm
45.0000, # left_lower_arm
45.0000, # right_thigh: x
135.0000, # right_thigh: y
45.0000, # right_thigh: z
45.0000, # left_thigh: x
135.0000, # left_thigh: y
45.0000, # left_thigh: z
90.0000, # right_knee
90.0000, # left_knee
22.5, # right_foot
22.5, # right_foot
22.5, # left_foot
22.5, # left_foot
],
device=self._device,
)
self.max_motor_effort = torch.max(self.joint_gears)
self.motor_effort_ratio = self.joint_gears / self.max_motor_effort
dof_limits = self._humanoids.get_dof_limits()
self.dof_limits_lower = dof_limits[0, :, 0].to(self._device)
self.dof_limits_upper = dof_limits[0, :, 1].to(self._device)
force_links = ["left_foot", "right_foot"]
self._sensor_indices = torch.tensor(
[self._humanoids._body_indices[j] for j in force_links], device=self._device, dtype=torch.long
)
LocomotionTask.post_reset(self)
def get_dof_at_limit_cost(self):
return get_dof_at_limit_cost(self.obs_buf, self.motor_effort_ratio, self.joints_at_limit_cost_scale)
@torch.jit.script
def get_dof_at_limit_cost(obs_buf, motor_effort_ratio, joints_at_limit_cost_scale):
# type: (Tensor, Tensor, float) -> Tensor
scaled_cost = joints_at_limit_cost_scale * (torch.abs(obs_buf[:, 12:33]) - 0.98) / 0.02
dof_at_limit_cost = torch.sum(
(torch.abs(obs_buf[:, 12:33]) > 0.98) * scaled_cost * motor_effort_ratio.unsqueeze(0), dim=-1
)
return dof_at_limit_cost
| 5,980 | Python | 41.119718 | 117 | 0.651003 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/franka_deformable.py | # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
#
# NVIDIA CORPORATION and its licensors retain all intellectual property
# and proprietary rights in and to this software, related documentation
# and any modifications thereto. Any use, reproduction, disclosure or
# distribution of this software and related documentation without an express
# license agreement from NVIDIA CORPORATION is strictly prohibited.
#
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.franka import Franka
from omniisaacgymenvs.robots.articulations.views.franka_view import FrankaView
from omni.isaac.core.prims import RigidPrim, RigidPrimView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage, add_reference_to_stage
from omni.isaac.core.utils.nucleus import get_assets_root_path
from omni.isaac.core.utils.torch.transformations import *
from omni.isaac.core.utils.torch.rotations import *
import omni.isaac.core.utils.deformable_mesh_utils as deformableMeshUtils
from omni.isaac.core.materials.deformable_material import DeformableMaterial
from omni.isaac.core.prims.soft.deformable_prim import DeformablePrim
from omni.isaac.core.prims.soft.deformable_prim_view import DeformablePrimView
from omni.physx.scripts import deformableUtils, physicsUtils
import numpy as np
import torch
import math
from pxr import Usd, UsdGeom, Gf, UsdPhysics, PhysxSchema
class FrankaDeformableTask(RLTask):
def __init__(
self,
name,
sim_config,
env,
offset=None
) -> None:
self.update_config(sim_config)
self.dt = 1/60.
self._num_observations = 39
self._num_actions = 9
RLTask.__init__(self, name, env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["episodeLength"]
self.dof_vel_scale = self._task_cfg["env"]["dofVelocityScale"]
self.action_scale = self._task_cfg["env"]["actionScale"]
def set_up_scene(self, scene) -> None:
self.stage = get_current_stage()
self.assets_root_path = get_assets_root_path()
if self.assets_root_path is None:
carb.log_error("Could not find Isaac Sim assets folder")
self.get_franka()
self.get_beaker()
self.get_deformable_tube()
super().set_up_scene(scene=scene, replicate_physics=False)
self._frankas = FrankaView(prim_paths_expr="/World/envs/.*/franka", name="franka_view")
self.deformableView = DeformablePrimView(
prim_paths_expr="/World/envs/.*/deformableTube/tube/mesh", name="deformabletube_view"
)
scene.add(self.deformableView)
scene.add(self._frankas)
scene.add(self._frankas._hands)
scene.add(self._frankas._lfingers)
scene.add(self._frankas._rfingers)
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("franka_view"):
scene.remove_object("franka_view", registry_only=True)
if scene.object_exists("hands_view"):
scene.remove_object("hands_view", registry_only=True)
if scene.object_exists("lfingers_view"):
scene.remove_object("lfingers_view", registry_only=True)
if scene.object_exists("rfingers_view"):
scene.remove_object("rfingers_view", registry_only=True)
if scene.object_exists("deformabletube_view"):
scene.remove_object("deformabletube_view", registry_only=True)
self._frankas = FrankaView(
prim_paths_expr="/World/envs/.*/franka", name="franka_view"
)
self.deformableView = DeformablePrimView(
prim_paths_expr="/World/envs/.*/deformableTube/tube/mesh", name="deformabletube_view"
)
scene.add(self._frankas)
scene.add(self._frankas._hands)
scene.add(self._frankas._lfingers)
scene.add(self._frankas._rfingers)
scene.add(self.deformableView)
def get_franka(self):
franka = Franka(
prim_path=self.default_zero_env_path + "/franka",
name="franka",
orientation=torch.tensor([1.0, 0.0, 0.0, 0.0]),
translation=torch.tensor([0.0, 0.0, 0.0]),
)
self._sim_config.apply_articulation_settings(
"franka", get_prim_at_path(franka.prim_path), self._sim_config.parse_actor_config("franka")
)
franka.set_franka_properties(stage=self.stage, prim=franka.prim)
def get_beaker(self):
_usd_path = self.assets_root_path + "/Isaac/Props/Beaker/beaker_500ml.usd"
mesh_path = self.default_zero_env_path + "/beaker"
add_reference_to_stage(_usd_path, mesh_path)
beaker = RigidPrim(
prim_path=mesh_path+"/beaker",
name="beaker",
position=torch.tensor([0.5, 0.2, 0.095]),
)
self._sim_config.apply_articulation_settings("beaker", beaker.prim, self._sim_config.parse_actor_config("beaker"))
def get_deformable_tube(self):
_usd_path = self.assets_root_path + "/Isaac/Props/DeformableTube/tube.usd"
mesh_path = self.default_zero_env_path + "/deformableTube/tube"
add_reference_to_stage(_usd_path, mesh_path)
skin_mesh = get_prim_at_path(mesh_path)
physicsUtils.setup_transform_as_scale_orient_translate(skin_mesh)
physicsUtils.set_or_add_translate_op(skin_mesh, (0.6, 0.0, 0.005))
physicsUtils.set_or_add_orient_op(skin_mesh, Gf.Rotation(Gf.Vec3d([0, 0, 1]), 90).GetQuat())
def get_observations(self) -> dict:
franka_dof_pos = self._frankas.get_joint_positions(clone=False)
franka_dof_vel = self._frankas.get_joint_velocities(clone=False)
self.franka_dof_pos = franka_dof_pos
dof_pos_scaled = (
2.0 * (franka_dof_pos - self.franka_dof_lower_limits)
/ (self.franka_dof_upper_limits - self.franka_dof_lower_limits)
- 1.0
)
self.lfinger_pos, _ = self._frankas._lfingers.get_world_poses(clone=False)
self.rfinger_pos, _ = self._frankas._rfingers.get_world_poses(clone=False)
self.gripper_site_pos = (self.lfinger_pos + self.rfinger_pos)/2 - self._env_pos
tube_positions = self.deformableView.get_simulation_mesh_nodal_positions(clone=False)
tube_velocities = self.deformableView.get_simulation_mesh_nodal_velocities(clone=False)
self.tube_front_positions = tube_positions[:, 200, :] - self._env_pos
self.tube_front_velocities = tube_velocities[:, 200, :]
self.tube_back_positions = tube_positions[:, -1, :] - self._env_pos
self.tube_back_velocities = tube_velocities[:, -1, :]
front_to_gripper = self.tube_front_positions - self.gripper_site_pos
to_front_goal = self.front_goal_pos - self.tube_front_positions
to_back_goal = self.back_goal_pos - self.tube_back_positions
self.obs_buf = torch.cat(
(
dof_pos_scaled,
franka_dof_vel * self.dof_vel_scale,
front_to_gripper,
to_front_goal,
to_back_goal,
self.tube_front_positions,
self.tube_front_velocities,
self.tube_back_positions,
self.tube_back_velocities,
),
dim=-1,
)
observations = {
self._frankas.name: {
"obs_buf": self.obs_buf
}
}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
self.actions = actions.clone().to(self._device)
targets = self.franka_dof_targets + self.franka_dof_speed_scales * self.dt * self.actions * self.action_scale
self.franka_dof_targets[:] = tensor_clamp(targets, self.franka_dof_lower_limits, self.franka_dof_upper_limits)
self.franka_dof_targets[:, -1] = self.franka_dof_targets[:, -2]
env_ids_int32 = torch.arange(self._frankas.count, dtype=torch.int32, device=self._device)
self._frankas.set_joint_position_targets(self.franka_dof_targets, indices=env_ids_int32)
def reset_idx(self, env_ids):
indices = env_ids.to(dtype=torch.int32)
num_indices = len(indices)
pos = self.franka_default_dof_pos
dof_pos = torch.zeros((num_indices, self._frankas.num_dof), device=self._device)
dof_vel = torch.zeros((num_indices, self._frankas.num_dof), device=self._device)
dof_pos[:, :] = pos
self.franka_dof_targets[env_ids, :] = pos
self.franka_dof_pos[env_ids, :] = pos
self._frankas.set_joint_position_targets(self.franka_dof_targets[env_ids], indices=indices)
self._frankas.set_joint_positions(dof_pos, indices=indices)
self._frankas.set_joint_velocities(dof_vel, indices=indices)
self.deformableView.set_simulation_mesh_nodal_positions(self.initial_tube_positions[env_ids], indices)
self.deformableView.set_simulation_mesh_nodal_velocities(self.initial_tube_velocities[env_ids], indices)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def post_reset(self):
self.franka_default_dof_pos = torch.tensor(
[0.00, 0.63, 0.00, -2.15, 0.00, 2.76, 0.75, 0.02, 0.02], device=self._device
)
self.actions = torch.zeros((self._num_envs, self.num_actions), device=self._device)
self.front_goal_pos = torch.tensor([0.36, 0.0, 0.23], device=self._device).repeat((self._num_envs, 1))
self.back_goal_pos = torch.tensor([0.5, 0.2, 0.0], device=self._device).repeat((self._num_envs, 1))
self.goal_hand_rot = torch.tensor([0.0, 1.0, 0.0, 0.0], device=self._device).repeat((self.num_envs, 1))
self.lfinger_pos, _ = self._frankas._lfingers.get_world_poses(clone=False)
self.rfinger_pos, _ = self._frankas._rfingers.get_world_poses(clone=False)
self.gripper_site_pos = (self.lfinger_pos + self.rfinger_pos)/2 - self._env_pos
self.initial_tube_positions = self.deformableView.get_simulation_mesh_nodal_positions()
self.initial_tube_velocities = self.deformableView.get_simulation_mesh_nodal_velocities()
self.tube_front_positions = self.initial_tube_positions[:, 0, :] - self._env_pos
self.tube_front_velocities = self.initial_tube_velocities[:, 0, :]
self.tube_back_positions = self.initial_tube_positions[:, -1, :] - self._env_pos
self.tube_back_velocities = self.initial_tube_velocities[:, -1, :]
self.num_franka_dofs = self._frankas.num_dof
self.franka_dof_pos = torch.zeros((self.num_envs, self.num_franka_dofs), device=self._device)
dof_limits = self._frankas.get_dof_limits()
self.franka_dof_lower_limits = dof_limits[0, :, 0].to(device=self._device)
self.franka_dof_upper_limits = dof_limits[0, :, 1].to(device=self._device)
self.franka_dof_speed_scales = torch.ones_like(self.franka_dof_lower_limits)
self.franka_dof_speed_scales[self._frankas.gripper_indices] = 0.1
self.franka_dof_targets = torch.zeros(
(self._num_envs, self.num_franka_dofs), dtype=torch.float, device=self._device
)
# randomize all envs
indices = torch.arange(self._num_envs, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
def calculate_metrics(self) -> None:
goal_distance_error = torch.norm(self.tube_back_positions[:, 0:2] - self.back_goal_pos[:, 0:2], p = 2, dim = -1)
goal_dist_reward = 1.0 / (5*goal_distance_error + .025)
current_z_level = self.tube_back_positions[:, 2:3]
z_lift_level = torch.where(
goal_distance_error < 0.07, torch.zeros_like(current_z_level), torch.ones_like(current_z_level)*0.18
)
front_lift_error = torch.norm(current_z_level - z_lift_level, p = 2, dim = -1)
front_lift_reward = 1.0 / (5*front_lift_error + .025)
rewards = goal_dist_reward + 4*front_lift_reward
self.rew_buf[:] = rewards
def is_done(self) -> None:
self.reset_buf = torch.where(self.progress_buf >= self._max_episode_length - 1, torch.ones_like(self.reset_buf), self.reset_buf)
self.reset_buf = torch.where(self.tube_front_positions[:, 0] < 0, torch.ones_like(self.reset_buf), self.reset_buf)
self.reset_buf = torch.where(self.tube_front_positions[:, 0] > 1.0, torch.ones_like(self.reset_buf), self.reset_buf)
self.reset_buf = torch.where(self.tube_front_positions[:, 1] < -1.0, torch.ones_like(self.reset_buf), self.reset_buf)
self.reset_buf = torch.where(self.tube_front_positions[:, 1] > 1.0, torch.ones_like(self.reset_buf), self.reset_buf)
| 13,322 | Python | 42.825658 | 136 | 0.641045 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/ant.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch.maths import tensor_clamp, torch_rand_float, unscale
from omni.isaac.core.utils.torch.rotations import compute_heading_and_up, compute_rot, quat_conjugate
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.ant import Ant
from omniisaacgymenvs.tasks.shared.locomotion import LocomotionTask
from pxr import PhysxSchema
class AntLocomotionTask(LocomotionTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
LocomotionTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_observations = 60
self._num_actions = 8
self._ant_positions = torch.tensor([0, 0, 0.5])
LocomotionTask.update_config(self)
def set_up_scene(self, scene) -> None:
self.get_ant()
RLTask.set_up_scene(self, scene)
self._ants = ArticulationView(
prim_paths_expr="/World/envs/.*/Ant/torso", name="ant_view", reset_xform_properties=False
)
scene.add(self._ants)
return
def initialize_views(self, scene):
RLTask.initialize_views(self, scene)
if scene.object_exists("ant_view"):
scene.remove_object("ant_view", registry_only=True)
self._ants = ArticulationView(
prim_paths_expr="/World/envs/.*/Ant/torso", name="ant_view", reset_xform_properties=False
)
scene.add(self._ants)
def get_ant(self):
ant = Ant(prim_path=self.default_zero_env_path + "/Ant", name="Ant", translation=self._ant_positions)
self._sim_config.apply_articulation_settings(
"Ant", get_prim_at_path(ant.prim_path), self._sim_config.parse_actor_config("Ant")
)
def get_robot(self):
return self._ants
def post_reset(self):
self.joint_gears = torch.tensor([15, 15, 15, 15, 15, 15, 15, 15], dtype=torch.float32, device=self._device)
dof_limits = self._ants.get_dof_limits()
self.dof_limits_lower = dof_limits[0, :, 0].to(self._device)
self.dof_limits_upper = dof_limits[0, :, 1].to(self._device)
self.motor_effort_ratio = torch.ones_like(self.joint_gears, device=self._device)
force_links = ["front_left_foot", "front_right_foot", "left_back_foot", "right_back_foot"]
self._sensor_indices = torch.tensor(
[self._ants._body_indices[j] for j in force_links], device=self._device, dtype=torch.long
)
LocomotionTask.post_reset(self)
def get_dof_at_limit_cost(self):
return get_dof_at_limit_cost(self.obs_buf, self._ants.num_dof)
@torch.jit.script
def get_dof_at_limit_cost(obs_buf, num_dof):
# type: (Tensor, int) -> Tensor
return torch.sum(obs_buf[:, 12 : 12 + num_dof] > 0.99, dim=-1)
| 4,691 | Python | 41.654545 | 115 | 0.69708 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/cartpole.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.cartpole import Cartpole
class CartpoleTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._max_episode_length = 500
self._num_observations = 4
self._num_actions = 1
RLTask.__init__(self, name, env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._cartpole_positions = torch.tensor([0.0, 0.0, 2.0])
self._reset_dist = self._task_cfg["env"]["resetDist"]
self._max_push_effort = self._task_cfg["env"]["maxEffort"]
def set_up_scene(self, scene) -> None:
self.get_cartpole()
super().set_up_scene(scene)
self._cartpoles = ArticulationView(
prim_paths_expr="/World/envs/.*/Cartpole", name="cartpole_view", reset_xform_properties=False
)
scene.add(self._cartpoles)
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("cartpole_view"):
scene.remove_object("cartpole_view", registry_only=True)
self._cartpoles = ArticulationView(
prim_paths_expr="/World/envs/.*/Cartpole", name="cartpole_view", reset_xform_properties=False
)
scene.add(self._cartpoles)
def get_cartpole(self):
cartpole = Cartpole(
prim_path=self.default_zero_env_path + "/Cartpole", name="Cartpole", translation=self._cartpole_positions
)
# applies articulation settings from the task configuration yaml file
self._sim_config.apply_articulation_settings(
"Cartpole", get_prim_at_path(cartpole.prim_path), self._sim_config.parse_actor_config("Cartpole")
)
def get_observations(self) -> dict:
dof_pos = self._cartpoles.get_joint_positions(clone=False)
dof_vel = self._cartpoles.get_joint_velocities(clone=False)
self.cart_pos = dof_pos[:, self._cart_dof_idx]
self.cart_vel = dof_vel[:, self._cart_dof_idx]
self.pole_pos = dof_pos[:, self._pole_dof_idx]
self.pole_vel = dof_vel[:, self._pole_dof_idx]
self.obs_buf[:, 0] = self.cart_pos
self.obs_buf[:, 1] = self.cart_vel
self.obs_buf[:, 2] = self.pole_pos
self.obs_buf[:, 3] = self.pole_vel
observations = {self._cartpoles.name: {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
actions = actions.to(self._device)
forces = torch.zeros((self._cartpoles.count, self._cartpoles.num_dof), dtype=torch.float32, device=self._device)
forces[:, self._cart_dof_idx] = self._max_push_effort * actions[:, 0]
indices = torch.arange(self._cartpoles.count, dtype=torch.int32, device=self._device)
self._cartpoles.set_joint_efforts(forces, indices=indices)
def reset_idx(self, env_ids):
num_resets = len(env_ids)
# randomize DOF positions
dof_pos = torch.zeros((num_resets, self._cartpoles.num_dof), device=self._device)
dof_pos[:, self._cart_dof_idx] = 1.0 * (1.0 - 2.0 * torch.rand(num_resets, device=self._device))
dof_pos[:, self._pole_dof_idx] = 0.125 * math.pi * (1.0 - 2.0 * torch.rand(num_resets, device=self._device))
# randomize DOF velocities
dof_vel = torch.zeros((num_resets, self._cartpoles.num_dof), device=self._device)
dof_vel[:, self._cart_dof_idx] = 0.5 * (1.0 - 2.0 * torch.rand(num_resets, device=self._device))
dof_vel[:, self._pole_dof_idx] = 0.25 * math.pi * (1.0 - 2.0 * torch.rand(num_resets, device=self._device))
# apply resets
indices = env_ids.to(dtype=torch.int32)
self._cartpoles.set_joint_positions(dof_pos, indices=indices)
self._cartpoles.set_joint_velocities(dof_vel, indices=indices)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def post_reset(self):
self._cart_dof_idx = self._cartpoles.get_dof_index("cartJoint")
self._pole_dof_idx = self._cartpoles.get_dof_index("poleJoint")
# randomize all envs
indices = torch.arange(self._cartpoles.count, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
def calculate_metrics(self) -> None:
reward = 1.0 - self.pole_pos * self.pole_pos - 0.01 * torch.abs(self.cart_vel) - 0.005 * torch.abs(self.pole_vel)
reward = torch.where(torch.abs(self.cart_pos) > self._reset_dist, torch.ones_like(reward) * -2.0, reward)
reward = torch.where(torch.abs(self.pole_pos) > np.pi / 2, torch.ones_like(reward) * -2.0, reward)
self.rew_buf[:] = reward
def is_done(self) -> None:
resets = torch.where(torch.abs(self.cart_pos) > self._reset_dist, 1, 0)
resets = torch.where(torch.abs(self.pole_pos) > math.pi / 2, 1, resets)
resets = torch.where(self.progress_buf >= self._max_episode_length, 1, resets)
self.reset_buf[:] = resets
| 7,256 | Python | 42.981818 | 121 | 0.659179 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/kukakr120r2500pro_reacher.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# Copyright (c) 2022-2023, Johnson Sun
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
# Ref: /omniisaacgymenvs/tasks/shadow_hand.py
import math
import numpy as np
import torch
# from omniisaacgymenvs.sim2real.kukakr120r2500pro import RealWorldKukaKR120R2500Pro
from omniisaacgymenvs.utils.config_utils.sim_config import SimConfig
from omniisaacgymenvs.robots.articulations.views.kukakr120r2500pro_view import KukaKR120R2500ProView
from omniisaacgymenvs.robots.articulations.kukakr120r2500pro import KukaKR120R2500Pro
from omniisaacgymenvs.tasks.shared.reacher import ReacherTask
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch import *
from omni.isaac.gym.vec_env import VecEnvBase
class KukaKR120R2500ProReacherTask(ReacherTask):
def __init__(
self,
name: str,
sim_config: SimConfig,
env: VecEnvBase,
offset=None
) -> None:
self.update_config(sim_config)
ReacherTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self.obs_type = self._task_cfg["env"]["observationType"]
if not (self.obs_type in ["full"]):
raise Exception(
"Unknown type of observations!\nobservationType should be one of: [full]")
print("Obs type:", self.obs_type)
self.num_obs_dict = {
"full": 29,
# 6: Kuka joints position (action space)
# 6: Kuka joints velocity
# 3: goal position
# 4: goal rotation
# 4: goal relative rotation
# 6: previous action
}
self.object_scale = torch.tensor([1.0] * 3)
self.goal_scale = torch.tensor([2.0] * 3)
self._num_observations = self.num_obs_dict[self.obs_type]
self._num_actions = 6
self._num_states = 0
pi = math.pi
# For actions
# The dof limits follow those defined in:
# thirdparty/kuka_kr120_support/urdf/kr120r2500pro.urdf
self._dof_limits = torch.tensor([[
[np.deg2rad(-185), np.deg2rad(185)],
[np.deg2rad(-155), np.deg2rad(35)],
[np.deg2rad(-130), np.deg2rad(154)],
[np.deg2rad(-350), np.deg2rad(350)],
[np.deg2rad(-130), np.deg2rad(130)],
[np.deg2rad(-350), np.deg2rad(350)],
]], dtype=torch.float32, device=self._cfg["sim_device"])
self.useURDF = self._task_cfg["env"]["useURDF"]
# Setup Sim2Real
sim2real_config = self._task_cfg['sim2real']
if sim2real_config['enabled'] and self.test and self.num_envs == 1:
raise NotImplementedError("Sim2Real is not implemented for Kuka KR120R2500Pro yet!")
ReacherTask.update_config(self)
def get_num_dof(self):
return self._arms.num_dof
def get_arm(self):
if not self.useURDF:
usd_path = "omniverse://localhost/Projects/J3soon/Isaac/2023.1.0/Isaac/Robots/Kuka/KR120_R2500_Pro/kr120r2500pro_urdf_instanceable.usd"
else:
raise NotImplementedError("Only URDF is supported for Kuka KR120R2500Pro now!")
kuka = KukaKR120R2500Pro(
prim_path=self.default_zero_env_path + "/Kuka",
name="Kuka",
usd_path=usd_path
)
self._sim_config.apply_articulation_settings(
"kuka",
get_prim_at_path(kuka.prim_path),
self._sim_config.parse_actor_config("kuka"),
)
def get_arm_view(self, scene):
end_effector_prim_paths_expr = "/World/envs/.*/Kuka/tool0"
arm_view = KukaKR120R2500ProView(
prim_paths_expr="/World/envs/.*/Kuka",
end_effector_prim_paths_expr=end_effector_prim_paths_expr,
name="kuka_view"
)
scene.add(arm_view._end_effectors)
return arm_view
def get_object_displacement_tensor(self):
return torch.tensor([0.05, 0.0, 0.0], device=self.device).repeat((self.num_envs, 1))
def get_observations(self):
self.arm_dof_pos = self._arms.get_joint_positions()
self.arm_dof_vel = self._arms.get_joint_velocities()
if self.obs_type == "full_no_vel":
self.compute_full_observations(True)
elif self.obs_type == "full":
self.compute_full_observations()
else:
print("Unkown observations type!")
observations = {self._arms.name: {"obs_buf": self.obs_buf}}
return observations
def get_reset_target_new_pos(self, n_reset_envs):
# Randomly generate goal positions, although the resulting goal may still not be reachable.
new_pos = torch_rand_float(-1, 1, (n_reset_envs, 3), device=self.device)
new_pos[:, 0] = new_pos[:, 0] * 0.5 + 1.0 * torch.sign(new_pos[:, 0])
new_pos[:, 1] = new_pos[:, 1] * 0.5 + 1.0 * torch.sign(new_pos[:, 1])
new_pos[:, 2] = torch.abs(new_pos[:, 2] * 1.3) + 0.2
return new_pos
def compute_full_observations(self, no_vel=False):
if no_vel:
raise NotImplementedError()
else:
# There are many redundant information for the simple Reacher task, but we'll keep them for now.
self.obs_buf[:, 0:self.num_arm_dofs] = unscale(self.arm_dof_pos[:, :self.num_arm_dofs],
self.arm_dof_lower_limits, self.arm_dof_upper_limits)
self.obs_buf[:, self.num_arm_dofs:2*self.num_arm_dofs] = self.vel_obs_scale * self.arm_dof_vel[:, :self.num_arm_dofs]
base = 2 * self.num_arm_dofs
self.obs_buf[:, base+0:base+3] = self.goal_pos
self.obs_buf[:, base+3:base+7] = self.goal_rot
self.obs_buf[:, base+7:base+11] = quat_mul(self.object_rot, quat_conjugate(self.goal_rot))
self.obs_buf[:, base+11:base+17] = self.actions
def send_joint_pos(self, joint_pos):
# self.real_world_kukakr120r2500pro.send_joint_pos(joint_pos)
pass
| 7,583 | Python | 41.606741 | 147 | 0.64605 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/quadcopter.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.objects import DynamicSphere
from omni.isaac.core.prims import RigidPrimView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch.rotations import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.quadcopter import Quadcopter
from omniisaacgymenvs.robots.articulations.views.quadcopter_view import QuadcopterView
class QuadcopterTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._num_observations = 21
self._num_actions = 12
self._copter_position = torch.tensor([0, 0, 1.0])
RLTask.__init__(self, name=name, env=env)
max_thrust = 2.0
self.thrust_lower_limits = -max_thrust * torch.ones(4, device=self._device, dtype=torch.float32)
self.thrust_upper_limits = max_thrust * torch.ones(4, device=self._device, dtype=torch.float32)
self.all_indices = torch.arange(self._num_envs, dtype=torch.int32, device=self._device)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["maxEpisodeLength"]
self.dt = self._task_cfg["sim"]["dt"]
def set_up_scene(self, scene) -> None:
self.get_copter()
self.get_target()
RLTask.set_up_scene(self, scene)
self._copters = QuadcopterView(prim_paths_expr="/World/envs/.*/Quadcopter", name="quadcopter_view")
self._balls = RigidPrimView(
prim_paths_expr="/World/envs/.*/ball", name="targets_view", reset_xform_properties=False
)
self._balls._non_root_link = True # do not set states for kinematics
scene.add(self._copters)
scene.add(self._copters.rotors)
scene.add(self._balls)
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("quadcopter_view"):
scene.remove_object("quadcopter_view", registry_only=True)
if scene.object_exists("rotors_view"):
scene.remove_object("rotors_view", registry_only=True)
if scene.object_exists("targets_view"):
scene.remove_object("targets_view", registry_only=True)
self._copters = QuadcopterView(prim_paths_expr="/World/envs/.*/Quadcopter", name="quadcopter_view")
self._balls = RigidPrimView(
prim_paths_expr="/World/envs/.*/ball", name="targets_view", reset_xform_properties=False
)
scene.add(self._copters)
scene.add(self._copters.rotors)
scene.add(self._balls)
def get_copter(self):
copter = Quadcopter(
prim_path=self.default_zero_env_path + "/Quadcopter", name="quadcopter", translation=self._copter_position
)
self._sim_config.apply_articulation_settings(
"copter", get_prim_at_path(copter.prim_path), self._sim_config.parse_actor_config("copter")
)
def get_target(self):
radius = 0.05
color = torch.tensor([1, 0, 0])
ball = DynamicSphere(
prim_path=self.default_zero_env_path + "/ball",
name="target_0",
radius=radius,
color=color,
)
self._sim_config.apply_articulation_settings(
"ball", get_prim_at_path(ball.prim_path), self._sim_config.parse_actor_config("ball")
)
ball.set_collision_enabled(False)
def get_observations(self) -> dict:
self.root_pos, self.root_rot = self._copters.get_world_poses(clone=False)
self.root_velocities = self._copters.get_velocities(clone=False)
self.dof_pos = self._copters.get_joint_positions(clone=False)
root_positions = self.root_pos - self._env_pos
root_quats = self.root_rot
root_linvels = self.root_velocities[:, :3]
root_angvels = self.root_velocities[:, 3:]
self.obs_buf[..., 0:3] = (self.target_positions - root_positions) / 3
self.obs_buf[..., 3:7] = root_quats
self.obs_buf[..., 7:10] = root_linvels / 2
self.obs_buf[..., 10:13] = root_angvels / math.pi
self.obs_buf[..., 13:21] = self.dof_pos
observations = {self._copters.name: {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
actions = actions.clone().to(self._device)
dof_action_speed_scale = 8 * math.pi
self.dof_position_targets += self.dt * dof_action_speed_scale * actions[:, 0:8]
self.dof_position_targets[:] = tensor_clamp(
self.dof_position_targets, self.dof_lower_limits, self.dof_upper_limits
)
thrust_action_speed_scale = 100
self.thrusts += self.dt * thrust_action_speed_scale * actions[:, 8:12]
self.thrusts[:] = tensor_clamp(self.thrusts, self.thrust_lower_limits, self.thrust_upper_limits)
self.forces[:, 0, 2] = self.thrusts[:, 0]
self.forces[:, 1, 2] = self.thrusts[:, 1]
self.forces[:, 2, 2] = self.thrusts[:, 2]
self.forces[:, 3, 2] = self.thrusts[:, 3]
# clear actions for reset envs
self.thrusts[reset_env_ids] = 0.0
self.forces[reset_env_ids] = 0.0
self.dof_position_targets[reset_env_ids] = self.dof_pos[reset_env_ids]
# apply actions
self._copters.set_joint_position_targets(self.dof_position_targets)
self._copters.rotors.apply_forces(self.forces, is_global=False)
def post_reset(self):
# control tensors
self.dof_position_targets = torch.zeros(
(self._num_envs, self._copters.num_dof), dtype=torch.float32, device=self._device, requires_grad=False
)
self.thrusts = torch.zeros((self._num_envs, 4), dtype=torch.float32, device=self._device, requires_grad=False)
self.forces = torch.zeros(
(self._num_envs, self._copters.rotors.count // self._num_envs, 3),
dtype=torch.float32,
device=self._device,
requires_grad=False,
)
self.target_positions = torch.zeros((self._num_envs, 3), device=self._device)
self.target_positions[:, 2] = 1.0
self.root_pos, self.root_rot = self._copters.get_world_poses(clone=False)
self.root_velocities = self._copters.get_velocities(clone=False)
self.dof_pos = self._copters.get_joint_positions(clone=False)
self.dof_vel = self._copters.get_joint_velocities(clone=False)
self.initial_root_pos, self.initial_root_rot = self.root_pos.clone(), self.root_rot.clone()
dof_limits = self._copters.get_dof_limits()
self.dof_lower_limits = dof_limits[0][:, 0].to(device=self._device)
self.dof_upper_limits = dof_limits[0][:, 1].to(device=self._device)
def reset_idx(self, env_ids):
num_resets = len(env_ids)
self.dof_pos[env_ids, :] = torch_rand_float(-0.2, 0.2, (num_resets, self._copters.num_dof), device=self._device)
self.dof_vel[env_ids, :] = 0
root_pos = self.initial_root_pos.clone()
root_pos[env_ids, 0] += torch_rand_float(-1.5, 1.5, (num_resets, 1), device=self._device).view(-1)
root_pos[env_ids, 1] += torch_rand_float(-1.5, 1.5, (num_resets, 1), device=self._device).view(-1)
root_pos[env_ids, 2] += torch_rand_float(-0.2, 1.5, (num_resets, 1), device=self._device).view(-1)
root_velocities = self.root_velocities.clone()
root_velocities[env_ids] = 0
# apply resets
self._copters.set_joint_positions(self.dof_pos[env_ids], indices=env_ids)
self._copters.set_joint_velocities(self.dof_vel[env_ids], indices=env_ids)
self._copters.set_world_poses(root_pos[env_ids], self.initial_root_rot[env_ids].clone(), indices=env_ids)
self._copters.set_velocities(root_velocities[env_ids], indices=env_ids)
self._balls.set_world_poses(positions=self.target_positions[:, 0:3] + self._env_pos)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def calculate_metrics(self) -> None:
root_positions = self.root_pos - self._env_pos
root_quats = self.root_rot
root_angvels = self.root_velocities[:, 3:]
# distance to target
target_dist = torch.sqrt(torch.square(self.target_positions - root_positions).sum(-1))
pos_reward = 1.0 / (1.0 + 3 * target_dist * target_dist) # 2
self.target_dist = target_dist
self.root_positions = root_positions
# uprightness
ups = quat_axis(root_quats, 2)
tiltage = torch.abs(1 - ups[..., 2])
up_reward = 1.0 / (1.0 + 10 * tiltage * tiltage)
# spinning
spinnage = torch.abs(root_angvels[..., 2])
spinnage_reward = 1.0 / (1.0 + 0.001 * spinnage * spinnage)
rew = pos_reward + pos_reward * (up_reward + spinnage_reward + spinnage * spinnage * (-1 / 400))
rew = torch.clip(rew, 0.0, None)
self.rew_buf[:] = rew
def is_done(self) -> None:
# resets due to misbehavior
ones = torch.ones_like(self.reset_buf)
die = torch.zeros_like(self.reset_buf)
die = torch.where(self.target_dist > 3.0, ones, die)
die = torch.where(self.root_positions[..., 2] < 0.3, ones, die)
# resets due to episode length
self.reset_buf[:] = torch.where(self.progress_buf >= self._max_episode_length - 1, ones, die)
| 11,498 | Python | 42.889313 | 120 | 0.640633 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/ingenuity.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from omniisaacgymenvs.robots.articulations.ingenuity import Ingenuity
from omniisaacgymenvs.robots.articulations.views.ingenuity_view import IngenuityView
from omni.isaac.core.utils.torch.rotations import *
from omni.isaac.core.objects import DynamicSphere
from omni.isaac.core.prims import RigidPrimView
from omni.isaac.core.utils.prims import get_prim_at_path
from omniisaacgymenvs.tasks.base.rl_task import RLTask
import numpy as np
import torch
import math
class IngenuityTask(RLTask):
def __init__(
self,
name,
sim_config,
env,
offset=None
) -> None:
self.update_config(sim_config)
self.thrust_limit = 2000
self.thrust_lateral_component = 0.2
self._num_observations = 13
self._num_actions = 6
self._ingenuity_position = torch.tensor([0, 0, 1.0])
self._ball_position = torch.tensor([0, 0, 1.0])
RLTask.__init__(self, name=name, env=env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["maxEpisodeLength"]
self.dt = self._task_cfg["sim"]["dt"]
def set_up_scene(self, scene) -> None:
self.get_ingenuity()
self.get_target()
RLTask.set_up_scene(self, scene)
self._copters = IngenuityView(prim_paths_expr="/World/envs/.*/Ingenuity", name="ingenuity_view")
self._balls = RigidPrimView(prim_paths_expr="/World/envs/.*/ball", name="targets_view", reset_xform_properties=False)
self._balls._non_root_link = True # do not set states for kinematics
scene.add(self._copters)
scene.add(self._balls)
for i in range(2):
scene.add(self._copters.physics_rotors[i])
scene.add(self._copters.visual_rotors[i])
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("ingenuity_view"):
scene.remove_object("ingenuity_view", registry_only=True)
for i in range(2):
if scene.object_exists(f"physics_rotor_{i}_view"):
scene.remove_object(f"physics_rotor_{i}_view", registry_only=True)
if scene.object_exists(f"visual_rotor_{i}_view"):
scene.remove_object(f"visual_rotor_{i}_view", registry_only=True)
if scene.object_exists("targets_view"):
scene.remove_object("targets_view", registry_only=True)
self._copters = IngenuityView(prim_paths_expr="/World/envs/.*/Ingenuity", name="ingenuity_view")
self._balls = RigidPrimView(prim_paths_expr="/World/envs/.*/ball", name="targets_view", reset_xform_properties=False)
scene.add(self._copters)
scene.add(self._balls)
for i in range(2):
scene.add(self._copters.physics_rotors[i])
scene.add(self._copters.visual_rotors[i])
def get_ingenuity(self):
copter = Ingenuity(prim_path=self.default_zero_env_path + "/Ingenuity", name="ingenuity", translation=self._ingenuity_position)
self._sim_config.apply_articulation_settings("ingenuity", get_prim_at_path(copter.prim_path), self._sim_config.parse_actor_config("ingenuity"))
def get_target(self):
radius = 0.1
color = torch.tensor([1, 0, 0])
ball = DynamicSphere(
prim_path=self.default_zero_env_path + "/ball",
translation=self._ball_position,
name="target_0",
radius=radius,
color=color,
)
self._sim_config.apply_articulation_settings("ball", get_prim_at_path(ball.prim_path), self._sim_config.parse_actor_config("ball"))
ball.set_collision_enabled(False)
def get_observations(self) -> dict:
self.root_pos, self.root_rot = self._copters.get_world_poses(clone=False)
self.root_velocities = self._copters.get_velocities(clone=False)
root_positions = self.root_pos - self._env_pos
root_quats = self.root_rot
root_linvels = self.root_velocities[:, :3]
root_angvels = self.root_velocities[:, 3:]
self.obs_buf[..., 0:3] = (self.target_positions - root_positions) / 3
self.obs_buf[..., 3:7] = root_quats
self.obs_buf[..., 7:10] = root_linvels / 2
self.obs_buf[..., 10:13] = root_angvels / math.pi
observations = {
self._copters.name: {
"obs_buf": self.obs_buf
}
}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
set_target_ids = (self.progress_buf % 500 == 0).nonzero(as_tuple=False).squeeze(-1)
if len(set_target_ids) > 0:
self.set_targets(set_target_ids)
actions = actions.clone().to(self._device)
vertical_thrust_prop_0 = torch.clamp(actions[:, 2] * self.thrust_limit, -self.thrust_limit, self.thrust_limit)
vertical_thrust_prop_1 = torch.clamp(actions[:, 5] * self.thrust_limit, -self.thrust_limit, self.thrust_limit)
lateral_fraction_prop_0 = torch.clamp(
actions[:, 0:2] * self.thrust_lateral_component,
-self.thrust_lateral_component,
self.thrust_lateral_component,
)
lateral_fraction_prop_1 = torch.clamp(
actions[:, 3:5] * self.thrust_lateral_component,
-self.thrust_lateral_component,
self.thrust_lateral_component,
)
self.thrusts[:, 0, 2] = self.dt * vertical_thrust_prop_0
self.thrusts[:, 0, 0:2] = self.thrusts[:, 0, 2, None] * lateral_fraction_prop_0
self.thrusts[:, 1, 2] = self.dt * vertical_thrust_prop_1
self.thrusts[:, 1, 0:2] = self.thrusts[:, 1, 2, None] * lateral_fraction_prop_1
# clear actions for reset envs
self.thrusts[reset_env_ids] = 0
# spin spinning rotors
self.dof_vel[:, self.spinning_indices[0]] = 50
self.dof_vel[:, self.spinning_indices[1]] = -50
self._copters.set_joint_velocities(self.dof_vel)
# apply actions
for i in range(2):
self._copters.physics_rotors[i].apply_forces(self.thrusts[:, i], indices=self.all_indices)
def post_reset(self):
self.spinning_indices = torch.tensor([1, 3], device=self._device)
self.all_indices = torch.arange(self._num_envs, dtype=torch.int32, device=self._device)
self.target_positions = torch.zeros((self._num_envs, 3), device=self._device, dtype=torch.float32)
self.target_positions[:, 2] = 1
self.root_pos, self.root_rot = self._copters.get_world_poses()
self.root_velocities = self._copters.get_velocities()
self.dof_pos = self._copters.get_joint_positions()
self.dof_vel = self._copters.get_joint_velocities()
self.initial_ball_pos, self.initial_ball_rot = self._balls.get_world_poses()
self.initial_root_pos, self.initial_root_rot = self.root_pos.clone(), self.root_rot.clone()
# control tensors
self.thrusts = torch.zeros((self._num_envs, 2, 3), dtype=torch.float32, device=self._device)
def set_targets(self, env_ids):
num_sets = len(env_ids)
envs_long = env_ids.long()
# set target position randomly with x, y in (-1, 1) and z in (1, 2)
self.target_positions[envs_long, 0:2] = torch.rand((num_sets, 2), device=self._device) * 2 - 1
self.target_positions[envs_long, 2] = torch.rand(num_sets, device=self._device) + 1
# shift the target up so it visually aligns better
ball_pos = self.target_positions[envs_long] + self._env_pos[envs_long]
ball_pos[:, 2] += 0.4
self._balls.set_world_poses(ball_pos[:, 0:3], self.initial_ball_rot[envs_long].clone(), indices=env_ids)
def reset_idx(self, env_ids):
num_resets = len(env_ids)
self.dof_pos[env_ids, 1] = torch_rand_float(-0.2, 0.2, (num_resets, 1), device=self._device).squeeze()
self.dof_pos[env_ids, 3] = torch_rand_float(-0.2, 0.2, (num_resets, 1), device=self._device).squeeze()
self.dof_vel[env_ids, :] = 0
root_pos = self.initial_root_pos.clone()
root_pos[env_ids, 0] += torch_rand_float(-0.5, 0.5, (num_resets, 1), device=self._device).view(-1)
root_pos[env_ids, 1] += torch_rand_float(-0.5, 0.5, (num_resets, 1), device=self._device).view(-1)
root_pos[env_ids, 2] += torch_rand_float(-0.5, 0.5, (num_resets, 1), device=self._device).view(-1)
root_velocities = self.root_velocities.clone()
root_velocities[env_ids] = 0
# apply resets
self._copters.set_joint_positions(self.dof_pos[env_ids], indices=env_ids)
self._copters.set_joint_velocities(self.dof_vel[env_ids], indices=env_ids)
self._copters.set_world_poses(root_pos[env_ids], self.initial_root_rot[env_ids].clone(), indices=env_ids)
self._copters.set_velocities(root_velocities[env_ids], indices=env_ids)
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def calculate_metrics(self) -> None:
root_positions = self.root_pos - self._env_pos
root_quats = self.root_rot
root_angvels = self.root_velocities[:, 3:]
# distance to target
target_dist = torch.sqrt(torch.square(self.target_positions - root_positions).sum(-1))
pos_reward = 1.0 / (1.0 + 2.5 * target_dist * target_dist)
self.target_dist = target_dist
self.root_positions = root_positions
# uprightness
ups = quat_axis(root_quats, 2)
tiltage = torch.abs(1 - ups[..., 2])
up_reward = 1.0 / (1.0 + 30 * tiltage * tiltage)
# spinning
spinnage = torch.abs(root_angvels[..., 2])
spinnage_reward = 1.0 / (1.0 + 10 * spinnage * spinnage)
# combined reward
# uprightness and spinning only matter when close to the target
self.rew_buf[:] = pos_reward + pos_reward * (up_reward + spinnage_reward)
def is_done(self) -> None:
# resets due to misbehavior
ones = torch.ones_like(self.reset_buf)
die = torch.zeros_like(self.reset_buf)
die = torch.where(self.target_dist > 20.0, ones, die)
die = torch.where(self.root_positions[..., 2] < 0.5, ones, die)
# resets due to episode length
self.reset_buf[:] = torch.where(self.progress_buf >= self._max_episode_length - 1, ones, die)
| 12,391 | Python | 42.943262 | 151 | 0.635138 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/anymal.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch.rotations import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.anymal import Anymal
from omniisaacgymenvs.robots.articulations.views.anymal_view import AnymalView
from omniisaacgymenvs.tasks.utils.usd_utils import set_drive
class AnymalTask(RLTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
self.update_config(sim_config)
self._num_observations = 48
self._num_actions = 12
RLTask.__init__(self, name, env)
return
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
# normalization
self.lin_vel_scale = self._task_cfg["env"]["learn"]["linearVelocityScale"]
self.ang_vel_scale = self._task_cfg["env"]["learn"]["angularVelocityScale"]
self.dof_pos_scale = self._task_cfg["env"]["learn"]["dofPositionScale"]
self.dof_vel_scale = self._task_cfg["env"]["learn"]["dofVelocityScale"]
self.action_scale = self._task_cfg["env"]["control"]["actionScale"]
# reward scales
self.rew_scales = {}
self.rew_scales["lin_vel_xy"] = self._task_cfg["env"]["learn"]["linearVelocityXYRewardScale"]
self.rew_scales["ang_vel_z"] = self._task_cfg["env"]["learn"]["angularVelocityZRewardScale"]
self.rew_scales["lin_vel_z"] = self._task_cfg["env"]["learn"]["linearVelocityZRewardScale"]
self.rew_scales["joint_acc"] = self._task_cfg["env"]["learn"]["jointAccRewardScale"]
self.rew_scales["action_rate"] = self._task_cfg["env"]["learn"]["actionRateRewardScale"]
self.rew_scales["cosmetic"] = self._task_cfg["env"]["learn"]["cosmeticRewardScale"]
# command ranges
self.command_x_range = self._task_cfg["env"]["randomCommandVelocityRanges"]["linear_x"]
self.command_y_range = self._task_cfg["env"]["randomCommandVelocityRanges"]["linear_y"]
self.command_yaw_range = self._task_cfg["env"]["randomCommandVelocityRanges"]["yaw"]
# base init state
pos = self._task_cfg["env"]["baseInitState"]["pos"]
rot = self._task_cfg["env"]["baseInitState"]["rot"]
v_lin = self._task_cfg["env"]["baseInitState"]["vLinear"]
v_ang = self._task_cfg["env"]["baseInitState"]["vAngular"]
state = pos + rot + v_lin + v_ang
self.base_init_state = state
# default joint positions
self.named_default_joint_angles = self._task_cfg["env"]["defaultJointAngles"]
# other
self.dt = 1 / 60
self.max_episode_length_s = self._task_cfg["env"]["learn"]["episodeLength_s"]
self.max_episode_length = int(self.max_episode_length_s / self.dt + 0.5)
self.Kp = self._task_cfg["env"]["control"]["stiffness"]
self.Kd = self._task_cfg["env"]["control"]["damping"]
for key in self.rew_scales.keys():
self.rew_scales[key] *= self.dt
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._anymal_translation = torch.tensor([0.0, 0.0, 0.62])
self._env_spacing = self._task_cfg["env"]["envSpacing"]
def set_up_scene(self, scene) -> None:
self.get_anymal()
super().set_up_scene(scene)
self._anymals = AnymalView(prim_paths_expr="/World/envs/.*/anymal", name="anymalview")
scene.add(self._anymals)
scene.add(self._anymals._knees)
scene.add(self._anymals._base)
return
def initialize_views(self, scene):
super().initialize_views(scene)
if scene.object_exists("anymalview"):
scene.remove_object("anymalview", registry_only=True)
if scene.object_exists("knees_view"):
scene.remove_object("knees_view", registry_only=True)
if scene.object_exists("base_view"):
scene.remove_object("base_view", registry_only=True)
self._anymals = AnymalView(prim_paths_expr="/World/envs/.*/anymal", name="anymalview")
scene.add(self._anymals)
scene.add(self._anymals._knees)
scene.add(self._anymals._base)
def get_anymal(self):
anymal = Anymal(
prim_path=self.default_zero_env_path + "/anymal", name="Anymal", translation=self._anymal_translation
)
self._sim_config.apply_articulation_settings(
"Anymal", get_prim_at_path(anymal.prim_path), self._sim_config.parse_actor_config("Anymal")
)
# Configure joint properties
joint_paths = []
for quadrant in ["LF", "LH", "RF", "RH"]:
for component, abbrev in [("HIP", "H"), ("THIGH", "K")]:
joint_paths.append(f"{quadrant}_{component}/{quadrant}_{abbrev}FE")
joint_paths.append(f"base/{quadrant}_HAA")
for joint_path in joint_paths:
set_drive(f"{anymal.prim_path}/{joint_path}", "angular", "position", 0, 400, 40, 1000)
def get_observations(self) -> dict:
torso_position, torso_rotation = self._anymals.get_world_poses(clone=False)
root_velocities = self._anymals.get_velocities(clone=False)
dof_pos = self._anymals.get_joint_positions(clone=False)
dof_vel = self._anymals.get_joint_velocities(clone=False)
velocity = root_velocities[:, 0:3]
ang_velocity = root_velocities[:, 3:6]
base_lin_vel = quat_rotate_inverse(torso_rotation, velocity) * self.lin_vel_scale
base_ang_vel = quat_rotate_inverse(torso_rotation, ang_velocity) * self.ang_vel_scale
projected_gravity = quat_rotate(torso_rotation, self.gravity_vec)
dof_pos_scaled = (dof_pos - self.default_dof_pos) * self.dof_pos_scale
commands_scaled = self.commands * torch.tensor(
[self.lin_vel_scale, self.lin_vel_scale, self.ang_vel_scale],
requires_grad=False,
device=self.commands.device,
)
obs = torch.cat(
(
base_lin_vel,
base_ang_vel,
projected_gravity,
commands_scaled,
dof_pos_scaled,
dof_vel * self.dof_vel_scale,
self.actions,
),
dim=-1,
)
self.obs_buf[:] = obs
observations = {self._anymals.name: {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
indices = torch.arange(self._anymals.count, dtype=torch.int32, device=self._device)
self.actions[:] = actions.clone().to(self._device)
current_targets = self.current_targets + self.action_scale * self.actions * self.dt
self.current_targets[:] = tensor_clamp(
current_targets, self.anymal_dof_lower_limits, self.anymal_dof_upper_limits
)
self._anymals.set_joint_position_targets(self.current_targets, indices)
def reset_idx(self, env_ids):
num_resets = len(env_ids)
# randomize DOF velocities
velocities = torch_rand_float(-0.1, 0.1, (num_resets, self._anymals.num_dof), device=self._device)
dof_pos = self.default_dof_pos[env_ids]
dof_vel = velocities
self.current_targets[env_ids] = dof_pos[:]
root_vel = torch.zeros((num_resets, 6), device=self._device)
# apply resets
indices = env_ids.to(dtype=torch.int32)
self._anymals.set_joint_positions(dof_pos, indices)
self._anymals.set_joint_velocities(dof_vel, indices)
self._anymals.set_world_poses(
self.initial_root_pos[env_ids].clone(), self.initial_root_rot[env_ids].clone(), indices
)
self._anymals.set_velocities(root_vel, indices)
self.commands_x[env_ids] = torch_rand_float(
self.command_x_range[0], self.command_x_range[1], (num_resets, 1), device=self._device
).squeeze()
self.commands_y[env_ids] = torch_rand_float(
self.command_y_range[0], self.command_y_range[1], (num_resets, 1), device=self._device
).squeeze()
self.commands_yaw[env_ids] = torch_rand_float(
self.command_yaw_range[0], self.command_yaw_range[1], (num_resets, 1), device=self._device
).squeeze()
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
self.last_actions[env_ids] = 0.0
self.last_dof_vel[env_ids] = 0.0
def post_reset(self):
self.default_dof_pos = torch.zeros(
(self.num_envs, 12), dtype=torch.float, device=self.device, requires_grad=False
)
dof_names = self._anymals.dof_names
for i in range(self.num_actions):
name = dof_names[i]
angle = self.named_default_joint_angles[name]
self.default_dof_pos[:, i] = angle
self.initial_root_pos, self.initial_root_rot = self._anymals.get_world_poses()
self.current_targets = self.default_dof_pos.clone()
dof_limits = self._anymals.get_dof_limits()
self.anymal_dof_lower_limits = dof_limits[0, :, 0].to(device=self._device)
self.anymal_dof_upper_limits = dof_limits[0, :, 1].to(device=self._device)
self.commands = torch.zeros(self._num_envs, 3, dtype=torch.float, device=self._device, requires_grad=False)
self.commands_y = self.commands.view(self._num_envs, 3)[..., 1]
self.commands_x = self.commands.view(self._num_envs, 3)[..., 0]
self.commands_yaw = self.commands.view(self._num_envs, 3)[..., 2]
# initialize some data used later on
self.extras = {}
self.gravity_vec = torch.tensor([0.0, 0.0, -1.0], device=self._device).repeat((self._num_envs, 1))
self.actions = torch.zeros(
self._num_envs, self.num_actions, dtype=torch.float, device=self._device, requires_grad=False
)
self.last_dof_vel = torch.zeros(
(self._num_envs, 12), dtype=torch.float, device=self._device, requires_grad=False
)
self.last_actions = torch.zeros(
self._num_envs, self.num_actions, dtype=torch.float, device=self._device, requires_grad=False
)
self.time_out_buf = torch.zeros_like(self.reset_buf)
# randomize all envs
indices = torch.arange(self._anymals.count, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
def calculate_metrics(self) -> None:
torso_position, torso_rotation = self._anymals.get_world_poses(clone=False)
root_velocities = self._anymals.get_velocities(clone=False)
dof_pos = self._anymals.get_joint_positions(clone=False)
dof_vel = self._anymals.get_joint_velocities(clone=False)
velocity = root_velocities[:, 0:3]
ang_velocity = root_velocities[:, 3:6]
base_lin_vel = quat_rotate_inverse(torso_rotation, velocity)
base_ang_vel = quat_rotate_inverse(torso_rotation, ang_velocity)
# velocity tracking reward
lin_vel_error = torch.sum(torch.square(self.commands[:, :2] - base_lin_vel[:, :2]), dim=1)
ang_vel_error = torch.square(self.commands[:, 2] - base_ang_vel[:, 2])
rew_lin_vel_xy = torch.exp(-lin_vel_error / 0.25) * self.rew_scales["lin_vel_xy"]
rew_ang_vel_z = torch.exp(-ang_vel_error / 0.25) * self.rew_scales["ang_vel_z"]
rew_lin_vel_z = torch.square(base_lin_vel[:, 2]) * self.rew_scales["lin_vel_z"]
rew_joint_acc = torch.sum(torch.square(self.last_dof_vel - dof_vel), dim=1) * self.rew_scales["joint_acc"]
rew_action_rate = (
torch.sum(torch.square(self.last_actions - self.actions), dim=1) * self.rew_scales["action_rate"]
)
rew_cosmetic = (
torch.sum(torch.abs(dof_pos[:, 0:4] - self.default_dof_pos[:, 0:4]), dim=1) * self.rew_scales["cosmetic"]
)
total_reward = rew_lin_vel_xy + rew_ang_vel_z + rew_joint_acc + rew_action_rate + rew_cosmetic + rew_lin_vel_z
total_reward = torch.clip(total_reward, 0.0, None)
self.last_actions[:] = self.actions[:]
self.last_dof_vel[:] = dof_vel[:]
self.fallen_over = self._anymals.is_base_below_threshold(threshold=0.51, ground_heights=0.0)
total_reward[torch.nonzero(self.fallen_over)] = -1
self.rew_buf[:] = total_reward.detach()
def is_done(self) -> None:
# reset agents
time_out = self.progress_buf >= self.max_episode_length - 1
self.reset_buf[:] = time_out | self.fallen_over
| 14,350 | Python | 44.55873 | 118 | 0.630941 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/warp/humanoid.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from omniisaacgymenvs.tasks.warp.shared.locomotion import LocomotionTask
from omniisaacgymenvs.robots.articulations.humanoid import Humanoid
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omniisaacgymenvs.tasks.base.rl_task import RLTaskWarp
import numpy as np
import torch
import warp as wp
import math
class HumanoidLocomotionTask(LocomotionTask):
def __init__(
self,
name,
sim_config,
env,
offset=None
) -> None:
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_observations = 87
self._num_actions = 21
self._humanoid_positions = torch.tensor([0, 0, 1.34])
LocomotionTask.__init__(self, name=name, env=env)
return
def set_up_scene(self, scene) -> None:
self.get_humanoid()
RLTaskWarp.set_up_scene(self, scene)
self._humanoids = ArticulationView(prim_paths_expr="/World/envs/.*/Humanoid/torso", name="humanoid_view", reset_xform_properties=False)
scene.add(self._humanoids)
return
def get_humanoid(self):
humanoid = Humanoid(prim_path=self.default_zero_env_path + "/Humanoid", name="Humanoid", translation=self._humanoid_positions)
self._sim_config.apply_articulation_settings("Humanoid", get_prim_at_path(humanoid.prim_path),
self._sim_config.parse_actor_config("Humanoid"))
def get_robot(self):
return self._humanoids
def post_reset(self):
self.joint_gears = wp.array(
[
67.5000, # lower_waist
67.5000, # lower_waist
67.5000, # right_upper_arm
67.5000, # right_upper_arm
67.5000, # left_upper_arm
67.5000, # left_upper_arm
67.5000, # pelvis
45.0000, # right_lower_arm
45.0000, # left_lower_arm
45.0000, # right_thigh: x
135.0000, # right_thigh: y
45.0000, # right_thigh: z
45.0000, # left_thigh: x
135.0000, # left_thigh: y
45.0000, # left_thigh: z
90.0000, # right_knee
90.0000, # left_knee
22.5, # right_foot
22.5, # right_foot
22.5, # left_foot
22.5, # left_foot
],
device=self._device,
dtype=wp.float32
)
self.max_motor_effort = 135.0
self.motor_effort_ratio = wp.zeros(self._humanoids._num_dof, dtype=wp.float32, device=self._device)
wp.launch(compute_effort_ratio, dim=self._humanoids._num_dof,
inputs=[self.motor_effort_ratio, self.joint_gears, self.max_motor_effort], device=self._device)
dof_limits = self._humanoids.get_dof_limits().to(self._device)
self.dof_limits_lower = wp.zeros(self._humanoids._num_dof, dtype=wp.float32, device=self._device)
self.dof_limits_upper = wp.zeros(self._humanoids._num_dof, dtype=wp.float32, device=self._device)
wp.launch(parse_dof_limits, dim=self._humanoids._num_dof,
inputs=[self.dof_limits_lower, self.dof_limits_upper, dof_limits], device=self._device)
self.dof_at_limit_cost = wp.zeros(self._num_envs, dtype=wp.float32, device=self._device)
force_links = ["left_foot", "right_foot"]
self._sensor_indices = wp.array([self._humanoids._body_indices[j] for j in force_links], device=self._device, dtype=wp.int32)
LocomotionTask.post_reset(self)
def get_dof_at_limit_cost(self):
wp.launch(get_dof_at_limit_cost, dim=(self._num_envs, self._humanoids._num_dof),
inputs=[self.dof_at_limit_cost, self.obs_buf, self.motor_effort_ratio, self.joints_at_limit_cost_scale])
return self.dof_at_limit_cost
@wp.kernel
def compute_effort_ratio(motor_effort_ratio: wp.array(dtype=wp.float32),
joint_gears: wp.array(dtype=wp.float32),
max_motor_effort: float):
tid = wp.tid()
motor_effort_ratio[tid] = joint_gears[tid] / max_motor_effort
@wp.kernel
def parse_dof_limits(dof_limits_lower: wp.array(dtype=wp.float32),
dof_limits_upper: wp.array(dtype=wp.float32),
dof_limits: wp.array(dtype=wp.float32, ndim=3)):
tid = wp.tid()
dof_limits_lower[tid] = dof_limits[0, tid, 0]
dof_limits_upper[tid] = dof_limits[0, tid, 1]
@wp.kernel
def get_dof_at_limit_cost(dof_at_limit_cost: wp.array(dtype=wp.float32),
obs_buf: wp.array(dtype=wp.float32, ndim=2),
motor_effort_ratio: wp.array(dtype=wp.float32),
joints_at_limit_cost_scale: float):
i, j = wp.tid()
dof_i = j + 12
scaled_cost = joints_at_limit_cost_scale * (wp.abs(obs_buf[i, dof_i]) - 0.98) / 0.02
cost = 0.0
if wp.abs(obs_buf[i, dof_i]) > 0.98:
cost = scaled_cost * motor_effort_ratio[j]
dof_at_limit_cost[i] = cost
| 6,686 | Python | 42.422078 | 143 | 0.639994 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/warp/ant.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from omniisaacgymenvs.robots.articulations.ant import Ant
from omniisaacgymenvs.tasks.warp.shared.locomotion import LocomotionTask
from omni.isaac.core.utils.torch.rotations import compute_heading_and_up, compute_rot, quat_conjugate
from omni.isaac.core.utils.torch.maths import torch_rand_float, tensor_clamp, unscale
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omniisaacgymenvs.tasks.base.rl_task import RLTaskWarp
import numpy as np
import torch
import warp as wp
class AntLocomotionTask(LocomotionTask):
def __init__(
self,
name,
sim_config,
env,
offset=None
) -> None:
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_observations = 60
self._num_actions = 8
self._ant_positions = wp.array([0, 0, 0.5], dtype=wp.float32, device="cpu")
LocomotionTask.__init__(self, name=name, env=env)
return
def set_up_scene(self, scene) -> None:
self.get_ant()
RLTaskWarp.set_up_scene(self, scene)
self._ants = ArticulationView(prim_paths_expr="/World/envs/.*/Ant/torso", name="ant_view", reset_xform_properties=False)
scene.add(self._ants)
return
def get_ant(self):
ant = Ant(prim_path=self.default_zero_env_path + "/Ant", name="Ant", translation=self._ant_positions)
self._sim_config.apply_articulation_settings("Ant", get_prim_at_path(ant.prim_path), self._sim_config.parse_actor_config("Ant"))
def get_robot(self):
return self._ants
def post_reset(self):
self.joint_gears = wp.array([15, 15, 15, 15, 15, 15, 15, 15], dtype=wp.float32, device=self._device)
dof_limits = self._ants.get_dof_limits().to(self._device)
self.dof_limits_lower = wp.zeros(self._ants._num_dof, dtype=wp.float32, device=self._device)
self.dof_limits_upper = wp.zeros(self._ants._num_dof, dtype=wp.float32, device=self._device)
wp.launch(parse_dof_limits, dim=self._ants._num_dof,
inputs=[self.dof_limits_lower, self.dof_limits_upper, dof_limits], device=self._device)
self.motor_effort_ratio = wp.array([1, 1, 1, 1, 1, 1, 1, 1], dtype=wp.float32, device=self._device)
self.dof_at_limit_cost = wp.zeros(self._num_envs, dtype=wp.float32, device=self._device)
force_links = ["front_left_foot", "front_right_foot", "left_back_foot", "right_back_foot"]
self._sensor_indices = wp.array([self._ants._body_indices[j] for j in force_links], device=self._device, dtype=wp.int32)
LocomotionTask.post_reset(self)
def get_dof_at_limit_cost(self):
wp.launch(get_dof_at_limit_cost, dim=(self._num_envs, self._ants._num_dof),
inputs=[self.dof_at_limit_cost, self.obs_buf, self.motor_effort_ratio])
return self.dof_at_limit_cost
@wp.kernel
def get_dof_at_limit_cost(dof_at_limit_cost: wp.array(dtype=wp.float32),
obs_buf: wp.array(dtype=wp.float32, ndim=2),
motor_effort_ratio: wp.array(dtype=wp.float32)):
i, j = wp.tid()
dof_i = j + 12
cost = 0.0
if wp.abs(obs_buf[i, dof_i]) > 0.99:
cost = 1.0
dof_at_limit_cost[i] = cost
@wp.kernel
def parse_dof_limits(dof_limits_lower: wp.array(dtype=wp.float32),
dof_limits_upper: wp.array(dtype=wp.float32),
dof_limits: wp.array(dtype=wp.float32, ndim=3)):
tid = wp.tid()
dof_limits_lower[tid] = dof_limits[0, tid, 0]
dof_limits_upper[tid] = dof_limits[0, tid, 1] | 5,221 | Python | 44.807017 | 136 | 0.685309 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/warp/cartpole.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from omniisaacgymenvs.robots.articulations.cartpole import Cartpole
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
import omni.isaac.core.utils.warp as warp_utils
from omniisaacgymenvs.tasks.base.rl_task import RLTaskWarp
import numpy as np
import torch
import warp as wp
import math
class CartpoleTask(RLTaskWarp):
def __init__(
self,
name,
sim_config,
env,
offset=None
) -> None:
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._cartpole_positions = wp.array([0.0, 0.0, 2.0], dtype=wp.float32)
self._reset_dist = self._task_cfg["env"]["resetDist"]
self._max_push_effort = self._task_cfg["env"]["maxEffort"]
self._max_episode_length = 500
self._num_observations = 4
self._num_actions = 1
RLTaskWarp.__init__(self, name, env)
return
def set_up_scene(self, scene) -> None:
self.get_cartpole()
super().set_up_scene(scene)
self._cartpoles = ArticulationView(prim_paths_expr="/World/envs/.*/Cartpole", name="cartpole_view", reset_xform_properties=False)
scene.add(self._cartpoles)
return
def get_cartpole(self):
cartpole = Cartpole(prim_path=self.default_zero_env_path + "/Cartpole", name="Cartpole", translation=self._cartpole_positions)
# applies articulation settings from the task configuration yaml file
self._sim_config.apply_articulation_settings("Cartpole", get_prim_at_path(cartpole.prim_path), self._sim_config.parse_actor_config("Cartpole"))
def get_observations(self) -> dict:
dof_pos = self._cartpoles.get_joint_positions(clone=False)
dof_vel = self._cartpoles.get_joint_velocities(clone=False)
wp.launch(get_observations, dim=self._num_envs,
inputs=[self.obs_buf, dof_pos, dof_vel, self._cart_dof_idx, self._pole_dof_idx], device=self._device)
observations = {
self._cartpoles.name: {
"obs_buf": self.obs_buf
}
}
return observations
def pre_physics_step(self, actions) -> None:
self.reset_idx()
actions_wp = wp.from_torch(actions)
forces = wp.zeros((self._cartpoles.count, self._cartpoles.num_dof), dtype=wp.float32, device=self._device)
wp.launch(compute_forces, dim=self._num_envs,
inputs=[forces, actions_wp, self._cart_dof_idx, self._max_push_effort], device=self._device)
self._cartpoles.set_joint_efforts(forces)
def reset_idx(self):
reset_env_ids = wp.to_torch(self.reset_buf).nonzero(as_tuple=False).squeeze(-1)
num_resets = len(reset_env_ids)
indices = wp.from_torch(reset_env_ids.to(dtype=torch.int32), dtype=wp.int32)
if num_resets > 0:
wp.launch(reset_idx, num_resets,
inputs=[self.dof_pos, self.dof_vel, indices, self.reset_buf, self.progress_buf, self._cart_dof_idx, self._pole_dof_idx, self._rand_seed],
device=self._device)
# apply resets
self._cartpoles.set_joint_positions(self.dof_pos[indices], indices=indices)
self._cartpoles.set_joint_velocities(self.dof_vel[indices], indices=indices)
def post_reset(self):
self._cart_dof_idx = self._cartpoles.get_dof_index("cartJoint")
self._pole_dof_idx = self._cartpoles.get_dof_index("poleJoint")
self.dof_pos = wp.zeros((self._num_envs, self._cartpoles.num_dof), device=self._device, dtype=wp.float32)
self.dof_vel = wp.zeros((self._num_envs, self._cartpoles.num_dof), device=self._device, dtype=wp.float32)
# randomize all envs
self.reset_idx()
def calculate_metrics(self) -> None:
wp.launch(calculate_metrics, dim=self._num_envs,
inputs=[self.obs_buf, self.rew_buf, self._reset_dist], device=self._device)
def is_done(self) -> None:
wp.launch(is_done, dim=self._num_envs,
inputs=[self.obs_buf, self.reset_buf, self.progress_buf, self._reset_dist, self._max_episode_length],
device=self._device)
@wp.kernel
def reset_idx(dof_pos: wp.array(dtype=wp.float32, ndim=2),
dof_vel: wp.array(dtype=wp.float32, ndim=2),
indices: wp.array(dtype=wp.int32),
reset_buf: wp.array(dtype=wp.int32),
progress_buf: wp.array(dtype=wp.int32),
cart_dof_idx: int,
pole_dof_idx: int,
rand_seed: int):
i = wp.tid()
idx = indices[i]
rand_state = wp.rand_init(rand_seed, i)
# randomize DOF positions
dof_pos[idx, cart_dof_idx] = 1.0 * (1.0 - 2.0 * wp.randf(rand_state))
dof_pos[idx, pole_dof_idx] = 0.125 * warp_utils.PI * (1.0 - 2.0 * wp.randf(rand_state))
# randomize DOF velocities
dof_vel[idx, cart_dof_idx] = 0.5 * (1.0 - 2.0 * wp.randf(rand_state))
dof_vel[idx, pole_dof_idx] = 0.25 * warp_utils.PI * (1.0 - 2.0 * wp.randf(rand_state))
# bookkeeping
progress_buf[idx] = 0
reset_buf[idx] = 0
@wp.kernel
def compute_forces(forces: wp.array(dtype=wp.float32, ndim=2),
actions: wp.array(dtype=wp.float32, ndim=2),
cart_dof_idx: int,
max_push_effort: float):
i = wp.tid()
forces[i, cart_dof_idx] = max_push_effort * actions[i, 0]
@wp.kernel
def get_observations(obs_buf: wp.array(dtype=wp.float32, ndim=2),
dof_pos: wp.indexedarray(dtype=wp.float32, ndim=2),
dof_vel: wp.indexedarray(dtype=wp.float32, ndim=2),
cart_dof_idx: int,
pole_dof_idx: int):
i = wp.tid()
obs_buf[i, 0] = dof_pos[i, cart_dof_idx]
obs_buf[i, 1] = dof_vel[i, cart_dof_idx]
obs_buf[i, 2] = dof_pos[i, pole_dof_idx]
obs_buf[i, 3] = dof_vel[i, pole_dof_idx]
@wp.kernel
def calculate_metrics(obs_buf: wp.array(dtype=wp.float32, ndim=2),
rew_buf: wp.array(dtype=wp.float32),
reset_dist: float):
i = wp.tid()
cart_pos = obs_buf[i, 0]
cart_vel = obs_buf[i, 1]
pole_angle = obs_buf[i, 2]
pole_vel = obs_buf[i, 3]
rew_buf[i] = 1.0 - pole_angle * pole_angle - 0.01 * wp.abs(cart_vel) - 0.005 * wp.abs(pole_vel)
if wp.abs(cart_pos) > reset_dist or wp.abs(pole_angle) > warp_utils.PI / 2.0:
rew_buf[i] = -2.0
@wp.kernel
def is_done(obs_buf: wp.array(dtype=wp.float32, ndim=2),
reset_buf: wp.array(dtype=wp.int32),
progress_buf: wp.array(dtype=wp.int32),
reset_dist: float,
max_episode_length: int):
i = wp.tid()
cart_pos = obs_buf[i, 0]
pole_pos = obs_buf[i, 2]
if wp.abs(cart_pos) > reset_dist or wp.abs(pole_pos) > warp_utils.PI / 2.0 or progress_buf[i] > max_episode_length:
reset_buf[i] = 1
else:
reset_buf[i] = 0
| 8,665 | Python | 38.390909 | 154 | 0.635661 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/warp/shared/locomotion.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from abc import abstractmethod
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
import omni.isaac.core.utils.warp as warp_utils
from omniisaacgymenvs.tasks.base.rl_task import RLTaskWarp
import numpy as np
import torch
import warp as wp
class LocomotionTask(RLTaskWarp):
def __init__(
self,
name,
env,
offset=None
) -> None:
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["episodeLength"]
self.dof_vel_scale = self._task_cfg["env"]["dofVelocityScale"]
self.angular_velocity_scale = self._task_cfg["env"]["angularVelocityScale"]
self.contact_force_scale = self._task_cfg["env"]["contactForceScale"]
self.power_scale = self._task_cfg["env"]["powerScale"]
self.heading_weight = self._task_cfg["env"]["headingWeight"]
self.up_weight = self._task_cfg["env"]["upWeight"]
self.actions_cost_scale = self._task_cfg["env"]["actionsCost"]
self.energy_cost_scale = self._task_cfg["env"]["energyCost"]
self.joints_at_limit_cost_scale = self._task_cfg["env"]["jointsAtLimitCost"]
self.death_cost = self._task_cfg["env"]["deathCost"]
self.termination_height = self._task_cfg["env"]["terminationHeight"]
self.alive_reward_scale = self._task_cfg["env"]["alive_reward_scale"]
self._num_sensors = 2
RLTaskWarp.__init__(self, name, env)
return
@abstractmethod
def set_up_scene(self, scene) -> None:
pass
@abstractmethod
def get_robot(self):
pass
def get_observations(self) -> dict:
torso_position, torso_rotation = self._robots.get_world_poses(clone=False)
velocities = self._robots.get_velocities(clone=False)
dof_pos = self._robots.get_joint_positions(clone=False)
dof_vel = self._robots.get_joint_velocities(clone=False)
# force sensors attached to the feet
sensor_force_torques = self._robots.get_measured_joint_forces()
wp.launch(get_observations, dim=self._num_envs,
inputs=[self.obs_buf, torso_position, torso_rotation, self._env_pos, velocities, dof_pos, dof_vel,
self.prev_potentials, self.potentials, self.dt, self.target,
self.basis_vec0, self.basis_vec1, self.dof_limits_lower, self.dof_limits_upper, self.dof_vel_scale,
sensor_force_torques, self.contact_force_scale, self.actions, self.angular_velocity_scale,
self._robots._num_dof, self._num_sensors, self._sensor_indices], device=self._device
)
observations = {
self._robots.name: {
"obs_buf": self.obs_buf
}
}
return observations
def pre_physics_step(self, actions) -> None:
self.reset_idx()
actions_wp = wp.from_torch(actions)
self.actions = actions_wp
wp.launch(compute_forces, dim=(self._num_envs, self._robots._num_dof),
inputs=[self.forces, self.actions, self.joint_gears, self.power_scale], device=self._device)
# applies joint torques
self._robots.set_joint_efforts(self.forces)
def reset_idx(self):
reset_env_ids = wp.to_torch(self.reset_buf).nonzero(as_tuple=False).squeeze(-1)
num_resets = len(reset_env_ids)
indices = wp.from_torch(reset_env_ids.to(dtype=torch.int32), dtype=wp.int32)
if num_resets > 0:
wp.launch(reset_dofs, dim=(num_resets, self._robots._num_dof),
inputs=[self.dof_pos, self.dof_vel, self.initial_dof_pos, self.dof_limits_lower, self.dof_limits_upper, indices, self._rand_seed],
device=self._device)
wp.launch(reset_idx, dim=num_resets,
inputs=[self.root_pos, self.root_rot, self.initial_root_pos, self.initial_root_rot, self._env_pos,
self.target, self.prev_potentials, self.potentials, self.dt,
self.reset_buf, self.progress_buf, indices, self._rand_seed],
device=self._device)
# apply resets
self._robots.set_joint_positions(self.dof_pos[indices], indices=indices)
self._robots.set_joint_velocities(self.dof_vel[indices], indices=indices)
self._robots.set_world_poses(self.root_pos[indices], self.root_rot[indices], indices=indices)
self._robots.set_velocities(self.root_vel[indices], indices=indices)
def post_reset(self):
self._robots = self.get_robot()
self.initial_root_pos, self.initial_root_rot = self._robots.get_world_poses()
self.initial_dof_pos = self._robots.get_joint_positions()
# initialize some data used later on
self.basis_vec0 = wp.vec3(1, 0, 0)
self.basis_vec1 = wp.vec3(0, 0, 1)
self.target = wp.vec3(1000, 0, 0)
self.dt = 1.0 / 60.0
# initialize potentials
self.potentials = wp.zeros(self._num_envs, dtype=wp.float32, device=self._device)
self.prev_potentials = wp.zeros(self._num_envs, dtype=wp.float32, device=self._device)
wp.launch(init_potentials, dim=self._num_envs,
inputs=[self.potentials, self.prev_potentials, self.dt], device=self._device)
self.actions = wp.zeros((self.num_envs, self.num_actions), device=self._device, dtype=wp.float32)
self.forces = wp.zeros((self._num_envs, self._robots._num_dof), dtype=wp.float32, device=self._device)
self.dof_pos = wp.zeros((self.num_envs, self._robots._num_dof), device=self._device, dtype=wp.float32)
self.dof_vel = wp.zeros((self.num_envs, self._robots._num_dof), device=self._device, dtype=wp.float32)
self.root_pos = wp.zeros((self.num_envs, 3), device=self._device, dtype=wp.float32)
self.root_rot = wp.zeros((self.num_envs, 4), device=self._device, dtype=wp.float32)
self.root_vel = wp.zeros((self.num_envs, 6), device=self._device, dtype=wp.float32)
# randomize all env
self.reset_idx()
def calculate_metrics(self) -> None:
dof_at_limit_cost = self.get_dof_at_limit_cost()
wp.launch(calculate_metrics, dim=self._num_envs,
inputs=[self.rew_buf, self.obs_buf, self.actions, self.up_weight, self.heading_weight, self.potentials, self.prev_potentials,
self.actions_cost_scale, self.energy_cost_scale, self.termination_height,
self.death_cost, self._robots.num_dof, dof_at_limit_cost, self.alive_reward_scale, self.motor_effort_ratio],
device=self._device
)
def is_done(self) -> None:
wp.launch(is_done, dim=self._num_envs,
inputs=[self.obs_buf, self.termination_height, self.reset_buf, self.progress_buf, self._max_episode_length],
device=self._device
)
#####################################################################
###==========================warp kernels=========================###
#####################################################################
@wp.kernel
def init_potentials(potentials: wp.array(dtype=wp.float32),
prev_potentials: wp.array(dtype=wp.float32),
dt: float):
i = wp.tid()
potentials[i] = -1000.0 / dt
prev_potentials[i] = -1000.0 / dt
@wp.kernel
def reset_idx(root_pos: wp.array(dtype=wp.float32, ndim=2),
root_rot: wp.array(dtype=wp.float32, ndim=2),
initial_root_pos: wp.indexedarray(dtype=wp.float32, ndim=2),
initial_root_rot: wp.indexedarray(dtype=wp.float32, ndim=2),
env_pos: wp.array(dtype=wp.float32, ndim=2),
target: wp.vec3,
prev_potentials: wp.array(dtype=wp.float32),
potentials: wp.array(dtype=wp.float32),
dt: float,
reset_buf: wp.array(dtype=wp.int32),
progress_buf: wp.array(dtype=wp.int32),
indices: wp.array(dtype=wp.int32),
rand_seed: int):
i = wp.tid()
idx = indices[i]
# reset root states
for j in range(3):
root_pos[idx, j] = initial_root_pos[idx, j]
for j in range(4):
root_rot[idx, j] = initial_root_rot[idx, j]
# reset potentials
to_target = target - wp.vec3(initial_root_pos[idx, 0] - env_pos[idx, 0], initial_root_pos[idx, 1] - env_pos[idx, 1], target[2])
prev_potentials[idx] = -wp.length(to_target) / dt
potentials[idx] = -wp.length(to_target) / dt
temp = potentials[idx] - prev_potentials[idx]
# bookkeeping
reset_buf[idx] = 0
progress_buf[idx] = 0
@wp.kernel
def reset_dofs(dof_pos: wp.array(dtype=wp.float32, ndim=2),
dof_vel: wp.array(dtype=wp.float32, ndim=2),
initial_dof_pos: wp.indexedarray(dtype=wp.float32, ndim=2),
dof_limits_lower: wp.array(dtype=wp.float32),
dof_limits_upper: wp.array(dtype=wp.float32),
indices: wp.array(dtype=wp.int32),
rand_seed: int):
i, j = wp.tid()
idx = indices[i]
rand_state = wp.rand_init(rand_seed, i * j + j)
# randomize DOF positions and velocities
dof_pos[idx, j] = wp.clamp(wp.randf(rand_state, -0.2, 0.2) + initial_dof_pos[idx, j], dof_limits_lower[j], dof_limits_upper[j])
dof_vel[idx, j] = wp.randf(rand_state, -0.1, 0.1)
@wp.kernel
def compute_forces(forces: wp.array(dtype=wp.float32, ndim=2),
actions: wp.array(dtype=wp.float32, ndim=2),
joint_gears: wp.array(dtype=wp.float32),
power_scale: float):
i, j = wp.tid()
forces[i, j] = actions[i, j] * joint_gears[j] * power_scale
@wp.func
def get_euler_xyz(q: wp.quat):
qx = 0
qy = 1
qz = 2
qw = 3
# roll (x-axis rotation)
sinr_cosp = 2.0 * (q[qw] * q[qx] + q[qy] * q[qz])
cosr_cosp = q[qw] * q[qw] - q[qx] * q[qx] - q[qy] * q[qy] + q[qz] * q[qz]
roll = wp.atan2(sinr_cosp, cosr_cosp)
# pitch (y-axis rotation)
sinp = 2.0 * (q[qw] * q[qy] - q[qz] * q[qx])
if wp.abs(sinp) >= 1:
pitch = warp_utils.PI / 2.0 * (wp.abs(sinp)/sinp)
else:
pitch = wp.asin(sinp)
# yaw (z-axis rotation)
siny_cosp = 2.0 * (q[qw] * q[qz] + q[qx] * q[qy])
cosy_cosp = q[qw] * q[qw] + q[qx] * q[qx] - q[qy] * q[qy] - q[qz] * q[qz]
yaw = wp.atan2(siny_cosp, cosy_cosp)
rpy = wp.vec3(roll % (2.0 * warp_utils.PI), pitch % (2.0 * warp_utils.PI), yaw % (2.0 * warp_utils.PI))
return rpy
@wp.func
def compute_up_vec(torso_rotation: wp.quat, vec1: wp.vec3):
up_vec = wp.quat_rotate(torso_rotation, vec1)
return up_vec
@wp.func
def compute_heading_vec(torso_rotation: wp.quat, vec0: wp.vec3):
heading_vec = wp.quat_rotate(torso_rotation, vec0)
return heading_vec
@wp.func
def unscale(x:float, lower:float, upper:float):
return (2.0 * x - upper - lower) / (upper - lower)
@wp.func
def normalize_angle(x: float):
return wp.atan2(wp.sin(x), wp.cos(x))
@wp.kernel
def get_observations(
obs_buf: wp.array(dtype=wp.float32, ndim=2),
torso_pos: wp.indexedarray(dtype=wp.float32, ndim=2),
torso_rot: wp.indexedarray(dtype=wp.float32, ndim=2),
env_pos: wp.array(dtype=wp.float32, ndim=2),
velocity: wp.indexedarray(dtype=wp.float32, ndim=2),
dof_pos: wp.indexedarray(dtype=wp.float32, ndim=2),
dof_vel: wp.indexedarray(dtype=wp.float32, ndim=2),
prev_potentials: wp.array(dtype=wp.float32),
potentials: wp.array(dtype=wp.float32),
dt: float,
target: wp.vec3,
basis_vec0: wp.vec3,
basis_vec1: wp.vec3,
dof_limits_lower: wp.array(dtype=wp.float32),
dof_limits_upper: wp.array(dtype=wp.float32),
dof_vel_scale: float,
sensor_force_torques: wp.indexedarray(dtype=wp.float32, ndim=3),
contact_force_scale: float,
actions: wp.array(dtype=wp.float32, ndim=2),
angular_velocity_scale: float,
num_dofs: int,
num_sensors: int,
sensor_indices: wp.array(dtype=wp.int32)
):
i = wp.tid()
torso_position_x = torso_pos[i, 0] - env_pos[i, 0]
torso_position_y = torso_pos[i, 1] - env_pos[i, 1]
torso_position_z = torso_pos[i, 2] - env_pos[i, 2]
to_target = target - wp.vec3(torso_position_x, torso_position_y, target[2])
prev_potentials[i] = potentials[i]
potentials[i] = -wp.length(to_target) / dt
temp = potentials[i] - prev_potentials[i]
torso_quat = wp.quat(torso_rot[i, 1], torso_rot[i, 2], torso_rot[i, 3], torso_rot[i, 0])
up_vec = compute_up_vec(torso_quat, basis_vec1)
up_proj = up_vec[2]
heading_vec = compute_heading_vec(torso_quat, basis_vec0)
target_dir = wp.normalize(to_target)
heading_proj = wp.dot(heading_vec, target_dir)
lin_velocity = wp.vec3(velocity[i, 0], velocity[i, 1], velocity[i, 2])
ang_velocity = wp.vec3(velocity[i, 3], velocity[i, 4], velocity[i, 5])
rpy = get_euler_xyz(torso_quat)
vel_loc = wp.quat_rotate_inv(torso_quat, lin_velocity)
angvel_loc = wp.quat_rotate_inv(torso_quat, ang_velocity)
walk_target_angle = wp.atan2(target[2] - torso_position_z, target[0] - torso_position_x)
angle_to_target = walk_target_angle - rpy[2] # yaw
# obs_buf shapes: 1, 3, 3, 1, 1, 1, 1, 1, num_dofs, num_dofs, num_sensors * 6, num_dofs
obs_offset = 0
obs_buf[i, 0] = torso_position_z
obs_offset = obs_offset + 1
for j in range(3):
obs_buf[i, j+obs_offset] = vel_loc[j]
obs_offset = obs_offset + 3
for j in range(3):
obs_buf[i, j+obs_offset] = angvel_loc[j] * angular_velocity_scale
obs_offset = obs_offset + 3
obs_buf[i, obs_offset+0] = normalize_angle(rpy[2])
obs_buf[i, obs_offset+1] = normalize_angle(rpy[0])
obs_buf[i, obs_offset+2] = normalize_angle(angle_to_target)
obs_buf[i, obs_offset+3] = up_proj
obs_buf[i, obs_offset+4] = heading_proj
obs_offset = obs_offset + 5
for j in range(num_dofs):
obs_buf[i, obs_offset+j] = unscale(dof_pos[i, j], dof_limits_lower[j], dof_limits_upper[j])
obs_offset = obs_offset + num_dofs
for j in range(num_dofs):
obs_buf[i, obs_offset+j] = dof_vel[i, j] * dof_vel_scale
obs_offset = obs_offset + num_dofs
for j in range(num_sensors):
sensor_idx = sensor_indices[j]
for k in range(6):
obs_buf[i, obs_offset+j*6+k] = sensor_force_torques[i, sensor_idx, k] * contact_force_scale
obs_offset = obs_offset + (num_sensors * 6)
for j in range(num_dofs):
obs_buf[i, obs_offset+j] = actions[i, j]
@wp.kernel
def is_done(
obs_buf: wp.array(dtype=wp.float32, ndim=2),
termination_height: float,
reset_buf: wp.array(dtype=wp.int32),
progress_buf: wp.array(dtype=wp.int32),
max_episode_length: int
):
i = wp.tid()
if obs_buf[i, 0] < termination_height or progress_buf[i] >= max_episode_length - 1:
reset_buf[i] = 1
else:
reset_buf[i] = 0
@wp.kernel
def calculate_metrics(
rew_buf: wp.array(dtype=wp.float32),
obs_buf: wp.array(dtype=wp.float32, ndim=2),
actions: wp.array(dtype=wp.float32, ndim=2),
up_weight: float,
heading_weight: float,
potentials: wp.array(dtype=wp.float32),
prev_potentials: wp.array(dtype=wp.float32),
actions_cost_scale: float,
energy_cost_scale: float,
termination_height: float,
death_cost: float,
num_dof: int,
dof_at_limit_cost: wp.array(dtype=wp.float32),
alive_reward_scale: float,
motor_effort_ratio: wp.array(dtype=wp.float32)
):
i = wp.tid()
# heading reward
if obs_buf[i, 11] > 0.8:
heading_reward = heading_weight
else:
heading_reward = heading_weight * obs_buf[i, 11] / 0.8
# aligning up axis of robot and environment
up_reward = 0.0
if obs_buf[i, 10] > 0.93:
up_reward = up_weight
# energy penalty for movement
actions_cost = float(0.0)
electricity_cost = float(0.0)
for j in range(num_dof):
actions_cost = actions_cost + (actions[i, j] * actions[i, j])
electricity_cost = electricity_cost + (wp.abs(actions[i, j] * obs_buf[i, 12+num_dof+j]) * motor_effort_ratio[j])
# reward for duration of staying alive
progress_reward = potentials[i] - prev_potentials[i]
total_reward = (
progress_reward
+ alive_reward_scale
+ up_reward
+ heading_reward
- actions_cost_scale * actions_cost
- energy_cost_scale * electricity_cost
- dof_at_limit_cost[i]
)
# adjust reward for fallen agents
if obs_buf[i, 0] < termination_height:
total_reward = death_cost
rew_buf[i] = total_reward
| 18,233 | Python | 39.52 | 147 | 0.624198 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/base/rl_task.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import asyncio
from abc import abstractmethod
import numpy as np
import omni.isaac.core.utils.warp.tensor as wp_utils
import omni.kit
import omni.usd
import torch
import warp as wp
from gym import spaces
from omni.isaac.cloner import GridCloner
from omni.isaac.core.tasks import BaseTask
from omni.isaac.core.utils.prims import define_prim
from omni.isaac.core.utils.stage import get_current_stage
from omni.isaac.core.utils.types import ArticulationAction
from omni.isaac.gym.tasks.rl_task import RLTaskInterface
from omniisaacgymenvs.utils.domain_randomization.randomize import Randomizer
from pxr import Gf, UsdGeom, UsdLux
class RLTask(RLTaskInterface):
"""This class provides a PyTorch RL-specific interface for setting up RL tasks.
It includes utilities for setting up RL task related parameters,
cloning environments, and data collection for RL algorithms.
"""
def __init__(self, name, env, offset=None) -> None:
"""Initializes RL parameters, cloner object, and buffers.
Args:
name (str): name of the task.
env (VecEnvBase): an instance of the environment wrapper class to register task.
offset (Optional[np.ndarray], optional): offset applied to all assets of the task. Defaults to None.
"""
BaseTask.__init__(self, name=name, offset=offset)
self._rand_seed = self._cfg["seed"]
# optimization flags for pytorch JIT
torch._C._jit_set_nvfuser_enabled(False)
self.test = self._cfg["test"]
self._device = self._cfg["sim_device"]
# set up randomizer for DR
self._dr_randomizer = Randomizer(self._cfg, self._task_cfg)
if self._dr_randomizer.randomize:
import omni.replicator.isaac as dr
self.dr = dr
# set up replicator for camera data collection
if self._task_cfg["sim"].get("enable_cameras", False):
from omni.replicator.isaac.scripts.writers.pytorch_writer import PytorchWriter
from omni.replicator.isaac.scripts.writers.pytorch_listener import PytorchListener
import omni.replicator.core as rep
self.rep = rep
self.PytorchWriter = PytorchWriter
self.PytorchListener = PytorchListener
print("Task Device:", self._device)
self.randomize_actions = False
self.randomize_observations = False
self.clip_obs = self._task_cfg["env"].get("clipObservations", np.Inf)
self.clip_actions = self._task_cfg["env"].get("clipActions", np.Inf)
self.rl_device = self._cfg.get("rl_device", "cuda:0")
self.control_frequency_inv = self._task_cfg["env"].get("controlFrequencyInv", 1)
self.rendering_interval = self._task_cfg.get("renderingInterval", 1)
print("RL device: ", self.rl_device)
self._env = env
if not hasattr(self, "_num_agents"):
self._num_agents = 1 # used for multi-agent environments
if not hasattr(self, "_num_states"):
self._num_states = 0
# initialize data spaces (defaults to gym.Box)
if not hasattr(self, "action_space"):
self.action_space = spaces.Box(
np.ones(self.num_actions, dtype=np.float32) * -1.0, np.ones(self.num_actions, dtype=np.float32) * 1.0
)
if not hasattr(self, "observation_space"):
self.observation_space = spaces.Box(
np.ones(self.num_observations, dtype=np.float32) * -np.Inf,
np.ones(self.num_observations, dtype=np.float32) * np.Inf,
)
if not hasattr(self, "state_space"):
self.state_space = spaces.Box(
np.ones(self.num_states, dtype=np.float32) * -np.Inf,
np.ones(self.num_states, dtype=np.float32) * np.Inf,
)
self.cleanup()
def cleanup(self) -> None:
"""Prepares torch buffers for RL data collection."""
# prepare tensors
self.obs_buf = torch.zeros((self._num_envs, self.num_observations), device=self._device, dtype=torch.float)
self.states_buf = torch.zeros((self._num_envs, self.num_states), device=self._device, dtype=torch.float)
self.rew_buf = torch.zeros(self._num_envs, device=self._device, dtype=torch.float)
self.reset_buf = torch.ones(self._num_envs, device=self._device, dtype=torch.long)
self.progress_buf = torch.zeros(self._num_envs, device=self._device, dtype=torch.long)
self.extras = {}
def set_up_scene(
self, scene, replicate_physics=True, collision_filter_global_paths=[], filter_collisions=True, copy_from_source=False
) -> None:
"""Clones environments based on value provided in task config and applies collision filters to mask
collisions across environments.
Args:
scene (Scene): Scene to add objects to.
replicate_physics (bool): Clone physics using PhysX API for better performance.
collision_filter_global_paths (list): Prim paths of global objects that should not have collision masked.
filter_collisions (bool): Mask off collision between environments.
copy_from_source (bool): Copy from source prim when cloning instead of inheriting.
"""
super().set_up_scene(scene)
self._cloner = GridCloner(spacing=self._env_spacing)
self._cloner.define_base_env(self.default_base_env_path)
stage = omni.usd.get_context().get_stage()
UsdGeom.Xform.Define(stage, self.default_zero_env_path)
if self._task_cfg["sim"].get("add_ground_plane", True):
self._ground_plane_path = "/World/defaultGroundPlane"
collision_filter_global_paths.append(self._ground_plane_path)
scene.add_default_ground_plane(prim_path=self._ground_plane_path)
prim_paths = self._cloner.generate_paths("/World/envs/env", self._num_envs)
self._env_pos = self._cloner.clone(
source_prim_path="/World/envs/env_0", prim_paths=prim_paths, replicate_physics=replicate_physics, copy_from_source=copy_from_source
)
self._env_pos = torch.tensor(np.array(self._env_pos), device=self._device, dtype=torch.float)
if filter_collisions:
self._cloner.filter_collisions(
self._env._world.get_physics_context().prim_path,
"/World/collisions",
prim_paths,
collision_filter_global_paths,
)
if self._env._render:
self.set_initial_camera_params(camera_position=[10, 10, 3], camera_target=[0, 0, 0])
if self._task_cfg["sim"].get("add_distant_light", True):
self._create_distant_light()
def set_initial_camera_params(self, camera_position=[10, 10, 3], camera_target=[0, 0, 0]):
from omni.kit.viewport.utility import get_viewport_from_window_name
from omni.kit.viewport.utility.camera_state import ViewportCameraState
viewport_api_2 = get_viewport_from_window_name("Viewport")
viewport_api_2.set_active_camera("/OmniverseKit_Persp")
camera_state = ViewportCameraState("/OmniverseKit_Persp", viewport_api_2)
camera_state.set_position_world(Gf.Vec3d(camera_position[0], camera_position[1], camera_position[2]), True)
camera_state.set_target_world(Gf.Vec3d(camera_target[0], camera_target[1], camera_target[2]), True)
def _create_distant_light(self, prim_path="/World/defaultDistantLight", intensity=5000):
stage = get_current_stage()
light = UsdLux.DistantLight.Define(stage, prim_path)
light.CreateIntensityAttr().Set(intensity)
def initialize_views(self, scene):
"""Optionally implemented by individual task classes to initialize views used in the task.
This API is required for the extension workflow, where tasks are expected to train on a pre-defined stage.
Args:
scene (Scene): Scene to remove existing views and initialize/add new views.
"""
self._cloner = GridCloner(spacing=self._env_spacing)
pos, _ = self._cloner.get_clone_transforms(self._num_envs)
self._env_pos = torch.tensor(np.array(pos), device=self._device, dtype=torch.float)
@property
def default_base_env_path(self):
"""Retrieves default path to the parent of all env prims.
Returns:
default_base_env_path(str): Defaults to "/World/envs".
"""
return "/World/envs"
@property
def default_zero_env_path(self):
"""Retrieves default path to the first env prim (index 0).
Returns:
default_zero_env_path(str): Defaults to "/World/envs/env_0".
"""
return f"{self.default_base_env_path}/env_0"
def reset(self):
"""Flags all environments for reset."""
self.reset_buf = torch.ones_like(self.reset_buf)
def post_physics_step(self):
"""Processes RL required computations for observations, states, rewards, resets, and extras.
Also maintains progress buffer for tracking step count per environment.
Returns:
obs_buf(torch.Tensor): Tensor of observation data.
rew_buf(torch.Tensor): Tensor of rewards data.
reset_buf(torch.Tensor): Tensor of resets/dones data.
extras(dict): Dictionary of extras data.
"""
self.progress_buf[:] += 1
if self._env._world.is_playing():
self.get_observations()
self.get_states()
self.calculate_metrics()
self.is_done()
self.get_extras()
return self.obs_buf, self.rew_buf, self.reset_buf, self.extras
class RLTaskWarp(RLTask):
def cleanup(self) -> None:
"""Prepares torch buffers for RL data collection."""
# prepare tensors
self.obs_buf = wp.zeros((self._num_envs, self.num_observations), device=self._device, dtype=wp.float32)
self.states_buf = wp.zeros((self._num_envs, self.num_states), device=self._device, dtype=wp.float32)
self.rew_buf = wp.zeros(self._num_envs, device=self._device, dtype=wp.float32)
self.reset_buf = wp_utils.ones(self._num_envs, device=self._device, dtype=wp.int32)
self.progress_buf = wp.zeros(self._num_envs, device=self._device, dtype=wp.int32)
self.zero_states_buf_torch = torch.zeros(
(self._num_envs, self.num_states), device=self._device, dtype=torch.float32
)
self.extras = {}
def reset(self):
"""Flags all environments for reset."""
wp.launch(reset_progress, dim=self._num_envs, inputs=[self.progress_buf], device=self._device)
def post_physics_step(self):
"""Processes RL required computations for observations, states, rewards, resets, and extras.
Also maintains progress buffer for tracking step count per environment.
Returns:
obs_buf(torch.Tensor): Tensor of observation data.
rew_buf(torch.Tensor): Tensor of rewards data.
reset_buf(torch.Tensor): Tensor of resets/dones data.
extras(dict): Dictionary of extras data.
"""
wp.launch(increment_progress, dim=self._num_envs, inputs=[self.progress_buf], device=self._device)
if self._env._world.is_playing():
self.get_observations()
self.get_states()
self.calculate_metrics()
self.is_done()
self.get_extras()
obs_buf_torch = wp.to_torch(self.obs_buf)
rew_buf_torch = wp.to_torch(self.rew_buf)
reset_buf_torch = wp.to_torch(self.reset_buf)
return obs_buf_torch, rew_buf_torch, reset_buf_torch, self.extras
def get_states(self):
"""API for retrieving states buffer, used for asymmetric AC training.
Returns:
states_buf(torch.Tensor): States buffer.
"""
if self.num_states > 0:
return wp.to_torch(self.states_buf)
else:
return self.zero_states_buf_torch
def set_up_scene(self, scene) -> None:
"""Clones environments based on value provided in task config and applies collision filters to mask
collisions across environments.
Args:
scene (Scene): Scene to add objects to.
"""
super().set_up_scene(scene)
self._env_pos = wp.from_torch(self._env_pos)
@wp.kernel
def increment_progress(progress_buf: wp.array(dtype=wp.int32)):
i = wp.tid()
progress_buf[i] = progress_buf[i] + 1
@wp.kernel
def reset_progress(progress_buf: wp.array(dtype=wp.int32)):
i = wp.tid()
progress_buf[i] = 1
| 14,224 | Python | 41.717718 | 143 | 0.653121 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_base.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: base class.
Inherits Gym's RLTask class and abstract base class. Inherited by environment classes. Not directly executed.
Configuration defined in FactoryBase.yaml. Asset info defined in factory_asset_info_franka_table.yaml.
"""
import carb
import hydra
import math
import numpy as np
import torch
from omni.isaac.core.objects import FixedCuboid
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omniisaacgymenvs.robots.articulations.factory_franka import FactoryFranka
from pxr import PhysxSchema, UsdPhysics
import omniisaacgymenvs.tasks.factory.factory_control as fc
from omniisaacgymenvs.tasks.factory.factory_schema_class_base import FactoryABCBase
from omniisaacgymenvs.tasks.factory.factory_schema_config_base import (
FactorySchemaConfigBase,
)
class FactoryBase(RLTask, FactoryABCBase):
def __init__(self, name, sim_config, env) -> None:
"""Initialize instance variables. Initialize RLTask superclass."""
# Set instance variables from base YAML
self._get_base_yaml_params()
self._env_spacing = self.cfg_base.env.env_spacing
# Set instance variables from task and train YAMLs
self._sim_config = sim_config
self._cfg = sim_config.config # CL args, task config, and train config
self._task_cfg = sim_config.task_config # just task config
self._num_envs = sim_config.task_config["env"]["numEnvs"]
self._num_observations = sim_config.task_config["env"]["numObservations"]
self._num_actions = sim_config.task_config["env"]["numActions"]
super().__init__(name, env)
def _get_base_yaml_params(self):
"""Initialize instance variables from YAML files."""
cs = hydra.core.config_store.ConfigStore.instance()
cs.store(name="factory_schema_config_base", node=FactorySchemaConfigBase)
config_path = (
"task/FactoryBase.yaml" # relative to Gym's Hydra search path (cfg dir)
)
self.cfg_base = hydra.compose(config_name=config_path)
self.cfg_base = self.cfg_base["task"] # strip superfluous nesting
asset_info_path = "../tasks/factory/yaml/factory_asset_info_franka_table.yaml" # relative to Gym's Hydra search path (cfg dir)
self.asset_info_franka_table = hydra.compose(config_name=asset_info_path)
self.asset_info_franka_table = self.asset_info_franka_table[""][""][""][
"tasks"
]["factory"][
"yaml"
] # strip superfluous nesting
def import_franka_assets(self, add_to_stage=True):
"""Set Franka and table asset options. Import assets."""
self._stage = get_current_stage()
if add_to_stage:
franka_translation = np.array([self.cfg_base.env.franka_depth, 0.0, 0.0])
franka_orientation = np.array([0.0, 0.0, 0.0, 1.0])
franka = FactoryFranka(
prim_path=self.default_zero_env_path + "/franka",
name="franka",
translation=franka_translation,
orientation=franka_orientation,
)
self._sim_config.apply_articulation_settings(
"franka",
get_prim_at_path(franka.prim_path),
self._sim_config.parse_actor_config("franka"),
)
for link_prim in franka.prim.GetChildren():
if link_prim.HasAPI(PhysxSchema.PhysxRigidBodyAPI):
rb = PhysxSchema.PhysxRigidBodyAPI.Get(
self._stage, link_prim.GetPrimPath()
)
rb.GetDisableGravityAttr().Set(True)
rb.GetRetainAccelerationsAttr().Set(False)
if self.cfg_base.sim.add_damping:
rb.GetLinearDampingAttr().Set(
1.0
) # default = 0.0; increased to improve stability
rb.GetMaxLinearVelocityAttr().Set(
1.0
) # default = 1000.0; reduced to prevent CUDA errors
rb.GetAngularDampingAttr().Set(
5.0
) # default = 0.5; increased to improve stability
rb.GetMaxAngularVelocityAttr().Set(
2 / math.pi * 180
) # default = 64.0; reduced to prevent CUDA errors
else:
rb.GetLinearDampingAttr().Set(0.0)
rb.GetMaxLinearVelocityAttr().Set(1000.0)
rb.GetAngularDampingAttr().Set(0.5)
rb.GetMaxAngularVelocityAttr().Set(64 / math.pi * 180)
table_translation = np.array(
[0.0, 0.0, self.cfg_base.env.table_height * 0.5]
)
table_orientation = np.array([1.0, 0.0, 0.0, 0.0])
table = FixedCuboid(
prim_path=self.default_zero_env_path + "/table",
name="table",
translation=table_translation,
orientation=table_orientation,
scale=np.array(
[
self.asset_info_franka_table.table_depth,
self.asset_info_franka_table.table_width,
self.cfg_base.env.table_height,
]
),
size=1.0,
color=np.array([0, 0, 0]),
)
self.parse_controller_spec(add_to_stage=add_to_stage)
def acquire_base_tensors(self):
"""Acquire tensors."""
self.num_dofs = 9
self.env_pos = self._env_pos
self.dof_pos = torch.zeros((self.num_envs, self.num_dofs), device=self.device)
self.dof_vel = torch.zeros((self.num_envs, self.num_dofs), device=self.device)
self.dof_torque = torch.zeros(
(self.num_envs, self.num_dofs), device=self.device
)
self.fingertip_contact_wrench = torch.zeros(
(self.num_envs, 6), device=self.device
)
self.ctrl_target_fingertip_midpoint_pos = torch.zeros(
(self.num_envs, 3), device=self.device
)
self.ctrl_target_fingertip_midpoint_quat = torch.zeros(
(self.num_envs, 4), device=self.device
)
self.ctrl_target_dof_pos = torch.zeros(
(self.num_envs, self.num_dofs), device=self.device
)
self.ctrl_target_gripper_dof_pos = torch.zeros(
(self.num_envs, 2), device=self.device
)
self.ctrl_target_fingertip_contact_wrench = torch.zeros(
(self.num_envs, 6), device=self.device
)
self.prev_actions = torch.zeros(
(self.num_envs, self.num_actions), device=self.device
)
def refresh_base_tensors(self):
"""Refresh tensors."""
if not self._env._world.is_playing():
return
self.dof_pos = self.frankas.get_joint_positions(clone=False)
self.dof_vel = self.frankas.get_joint_velocities(clone=False)
# Jacobian shape: [4, 11, 6, 9] (root has no Jacobian)
self.franka_jacobian = self.frankas.get_jacobians()
self.franka_mass_matrix = self.frankas.get_mass_matrices(clone=False)
self.arm_dof_pos = self.dof_pos[:, 0:7]
self.arm_mass_matrix = self.franka_mass_matrix[
:, 0:7, 0:7
] # for Franka arm (not gripper)
self.hand_pos, self.hand_quat = self.frankas._hands.get_world_poses(clone=False)
self.hand_pos -= self.env_pos
hand_velocities = self.frankas._hands.get_velocities(clone=False)
self.hand_linvel = hand_velocities[:, 0:3]
self.hand_angvel = hand_velocities[:, 3:6]
(
self.left_finger_pos,
self.left_finger_quat,
) = self.frankas._lfingers.get_world_poses(clone=False)
self.left_finger_pos -= self.env_pos
left_finger_velocities = self.frankas._lfingers.get_velocities(clone=False)
self.left_finger_linvel = left_finger_velocities[:, 0:3]
self.left_finger_angvel = left_finger_velocities[:, 3:6]
self.left_finger_jacobian = self.franka_jacobian[:, 8, 0:6, 0:7]
left_finger_forces = self.frankas._lfingers.get_net_contact_forces(clone=False)
self.left_finger_force = left_finger_forces[:, 0:3]
(
self.right_finger_pos,
self.right_finger_quat,
) = self.frankas._rfingers.get_world_poses(clone=False)
self.right_finger_pos -= self.env_pos
right_finger_velocities = self.frankas._rfingers.get_velocities(clone=False)
self.right_finger_linvel = right_finger_velocities[:, 0:3]
self.right_finger_angvel = right_finger_velocities[:, 3:6]
self.right_finger_jacobian = self.franka_jacobian[:, 9, 0:6, 0:7]
right_finger_forces = self.frankas._rfingers.get_net_contact_forces(clone=False)
self.right_finger_force = right_finger_forces[:, 0:3]
self.gripper_dof_pos = self.dof_pos[:, 7:9]
(
self.fingertip_centered_pos,
self.fingertip_centered_quat,
) = self.frankas._fingertip_centered.get_world_poses(clone=False)
self.fingertip_centered_pos -= self.env_pos
fingertip_centered_velocities = self.frankas._fingertip_centered.get_velocities(
clone=False
)
self.fingertip_centered_linvel = fingertip_centered_velocities[:, 0:3]
self.fingertip_centered_angvel = fingertip_centered_velocities[:, 3:6]
self.fingertip_centered_jacobian = self.franka_jacobian[:, 10, 0:6, 0:7]
self.finger_midpoint_pos = (self.left_finger_pos + self.right_finger_pos) / 2
self.fingertip_midpoint_pos = fc.translate_along_local_z(
pos=self.finger_midpoint_pos,
quat=self.hand_quat,
offset=self.asset_info_franka_table.franka_finger_length,
device=self.device,
)
self.fingertip_midpoint_quat = self.fingertip_centered_quat # always equal
# TODO: Add relative velocity term (see https://dynamicsmotioncontrol487379916.files.wordpress.com/2020/11/21-me258pointmovingrigidbody.pdf)
self.fingertip_midpoint_linvel = self.fingertip_centered_linvel + torch.cross(
self.fingertip_centered_angvel,
(self.fingertip_midpoint_pos - self.fingertip_centered_pos),
dim=1,
)
# From sum of angular velocities (https://physics.stackexchange.com/questions/547698/understanding-addition-of-angular-velocity),
# angular velocity of midpoint w.r.t. world is equal to sum of
# angular velocity of midpoint w.r.t. hand and angular velocity of hand w.r.t. world.
# Midpoint is in sliding contact (i.e., linear relative motion) with hand; angular velocity of midpoint w.r.t. hand is zero.
# Thus, angular velocity of midpoint w.r.t. world is equal to angular velocity of hand w.r.t. world.
self.fingertip_midpoint_angvel = self.fingertip_centered_angvel # always equal
self.fingertip_midpoint_jacobian = (
self.left_finger_jacobian + self.right_finger_jacobian
) * 0.5
def parse_controller_spec(self, add_to_stage):
"""Parse controller specification into lower-level controller configuration."""
cfg_ctrl_keys = {
"num_envs",
"jacobian_type",
"gripper_prop_gains",
"gripper_deriv_gains",
"motor_ctrl_mode",
"gain_space",
"ik_method",
"joint_prop_gains",
"joint_deriv_gains",
"do_motion_ctrl",
"task_prop_gains",
"task_deriv_gains",
"do_inertial_comp",
"motion_ctrl_axes",
"do_force_ctrl",
"force_ctrl_method",
"wrench_prop_gains",
"force_ctrl_axes",
}
self.cfg_ctrl = {cfg_ctrl_key: None for cfg_ctrl_key in cfg_ctrl_keys}
self.cfg_ctrl["num_envs"] = self.num_envs
self.cfg_ctrl["jacobian_type"] = self.cfg_task.ctrl.all.jacobian_type
self.cfg_ctrl["gripper_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.all.gripper_prop_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["gripper_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.all.gripper_deriv_gains, device=self.device
).repeat((self.num_envs, 1))
ctrl_type = self.cfg_task.ctrl.ctrl_type
if ctrl_type == "gym_default":
self.cfg_ctrl["motor_ctrl_mode"] = "gym"
self.cfg_ctrl["gain_space"] = "joint"
self.cfg_ctrl["ik_method"] = self.cfg_task.ctrl.gym_default.ik_method
self.cfg_ctrl["joint_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.gym_default.joint_prop_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["joint_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.gym_default.joint_deriv_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["gripper_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.gym_default.gripper_prop_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["gripper_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.gym_default.gripper_deriv_gains, device=self.device
).repeat((self.num_envs, 1))
elif ctrl_type == "joint_space_ik":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "joint"
self.cfg_ctrl["ik_method"] = self.cfg_task.ctrl.joint_space_ik.ik_method
self.cfg_ctrl["joint_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.joint_space_ik.joint_prop_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["joint_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.joint_space_ik.joint_deriv_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_inertial_comp"] = False
elif ctrl_type == "joint_space_id":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "joint"
self.cfg_ctrl["ik_method"] = self.cfg_task.ctrl.joint_space_id.ik_method
self.cfg_ctrl["joint_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.joint_space_id.joint_prop_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["joint_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.joint_space_id.joint_deriv_gains, device=self.device
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_inertial_comp"] = True
elif ctrl_type == "task_space_impedance":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "task"
self.cfg_ctrl["do_motion_ctrl"] = True
self.cfg_ctrl["task_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.task_space_impedance.task_prop_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["task_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.task_space_impedance.task_deriv_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_inertial_comp"] = False
self.cfg_ctrl["motion_ctrl_axes"] = torch.tensor(
self.cfg_task.ctrl.task_space_impedance.motion_ctrl_axes,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_force_ctrl"] = False
elif ctrl_type == "operational_space_motion":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "task"
self.cfg_ctrl["do_motion_ctrl"] = True
self.cfg_ctrl["task_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.operational_space_motion.task_prop_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["task_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.operational_space_motion.task_deriv_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_inertial_comp"] = True
self.cfg_ctrl["motion_ctrl_axes"] = torch.tensor(
self.cfg_task.ctrl.operational_space_motion.motion_ctrl_axes,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_force_ctrl"] = False
elif ctrl_type == "open_loop_force":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "task"
self.cfg_ctrl["do_motion_ctrl"] = False
self.cfg_ctrl["do_force_ctrl"] = True
self.cfg_ctrl["force_ctrl_method"] = "open"
self.cfg_ctrl["force_ctrl_axes"] = torch.tensor(
self.cfg_task.ctrl.open_loop_force.force_ctrl_axes, device=self.device
).repeat((self.num_envs, 1))
elif ctrl_type == "closed_loop_force":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "task"
self.cfg_ctrl["do_motion_ctrl"] = False
self.cfg_ctrl["do_force_ctrl"] = True
self.cfg_ctrl["force_ctrl_method"] = "closed"
self.cfg_ctrl["wrench_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.closed_loop_force.wrench_prop_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["force_ctrl_axes"] = torch.tensor(
self.cfg_task.ctrl.closed_loop_force.force_ctrl_axes, device=self.device
).repeat((self.num_envs, 1))
elif ctrl_type == "hybrid_force_motion":
self.cfg_ctrl["motor_ctrl_mode"] = "manual"
self.cfg_ctrl["gain_space"] = "task"
self.cfg_ctrl["do_motion_ctrl"] = True
self.cfg_ctrl["task_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.hybrid_force_motion.task_prop_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["task_deriv_gains"] = torch.tensor(
self.cfg_task.ctrl.hybrid_force_motion.task_deriv_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_inertial_comp"] = True
self.cfg_ctrl["motion_ctrl_axes"] = torch.tensor(
self.cfg_task.ctrl.hybrid_force_motion.motion_ctrl_axes,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["do_force_ctrl"] = True
self.cfg_ctrl["force_ctrl_method"] = "closed"
self.cfg_ctrl["wrench_prop_gains"] = torch.tensor(
self.cfg_task.ctrl.hybrid_force_motion.wrench_prop_gains,
device=self.device,
).repeat((self.num_envs, 1))
self.cfg_ctrl["force_ctrl_axes"] = torch.tensor(
self.cfg_task.ctrl.hybrid_force_motion.force_ctrl_axes,
device=self.device,
).repeat((self.num_envs, 1))
if add_to_stage:
if self.cfg_ctrl["motor_ctrl_mode"] == "gym":
for i in range(7):
joint_prim = self._stage.GetPrimAtPath(
self.default_zero_env_path
+ f"/franka/panda_link{i}/panda_joint{i+1}"
)
drive = UsdPhysics.DriveAPI.Apply(joint_prim, "angular")
drive.GetStiffnessAttr().Set(
self.cfg_ctrl["joint_prop_gains"][0, i].item() * np.pi / 180
)
drive.GetDampingAttr().Set(
self.cfg_ctrl["joint_deriv_gains"][0, i].item() * np.pi / 180
)
for i in range(2):
joint_prim = self._stage.GetPrimAtPath(
self.default_zero_env_path
+ f"/franka/panda_hand/panda_finger_joint{i+1}"
)
drive = UsdPhysics.DriveAPI.Apply(joint_prim, "linear")
drive.GetStiffnessAttr().Set(
self.cfg_ctrl["gripper_deriv_gains"][0, i].item()
)
drive.GetDampingAttr().Set(
self.cfg_ctrl["gripper_deriv_gains"][0, i].item()
)
elif self.cfg_ctrl["motor_ctrl_mode"] == "manual":
for i in range(7):
joint_prim = self._stage.GetPrimAtPath(
self.default_zero_env_path
+ f"/franka/panda_link{i}/panda_joint{i+1}"
)
joint_prim.RemoveAPI(UsdPhysics.DriveAPI, "angular")
drive = UsdPhysics.DriveAPI.Apply(joint_prim, "None")
drive.GetStiffnessAttr().Set(0.0)
drive.GetDampingAttr().Set(0.0)
for i in range(2):
joint_prim = self._stage.GetPrimAtPath(
self.default_zero_env_path
+ f"/franka/panda_hand/panda_finger_joint{i+1}"
)
joint_prim.RemoveAPI(UsdPhysics.DriveAPI, "linear")
drive = UsdPhysics.DriveAPI.Apply(joint_prim, "None")
drive.GetStiffnessAttr().Set(0.0)
drive.GetDampingAttr().Set(0.0)
def generate_ctrl_signals(self):
"""Get Jacobian. Set Franka DOF position targets or DOF torques."""
# Get desired Jacobian
if self.cfg_ctrl["jacobian_type"] == "geometric":
self.fingertip_midpoint_jacobian_tf = self.fingertip_midpoint_jacobian
elif self.cfg_ctrl["jacobian_type"] == "analytic":
self.fingertip_midpoint_jacobian_tf = fc.get_analytic_jacobian(
fingertip_quat=self.fingertip_quat,
fingertip_jacobian=self.fingertip_midpoint_jacobian,
num_envs=self.num_envs,
device=self.device,
)
# Set PD joint pos target or joint torque
if self.cfg_ctrl["motor_ctrl_mode"] == "gym":
self._set_dof_pos_target()
elif self.cfg_ctrl["motor_ctrl_mode"] == "manual":
self._set_dof_torque()
def _set_dof_pos_target(self):
"""Set Franka DOF position target to move fingertips towards target pose."""
self.ctrl_target_dof_pos = fc.compute_dof_pos_target(
cfg_ctrl=self.cfg_ctrl,
arm_dof_pos=self.arm_dof_pos,
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
jacobian=self.fingertip_midpoint_jacobian_tf,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
ctrl_target_gripper_dof_pos=self.ctrl_target_gripper_dof_pos,
device=self.device,
)
self.frankas.set_joint_position_targets(positions=self.ctrl_target_dof_pos)
def _set_dof_torque(self):
"""Set Franka DOF torque to move fingertips towards target pose."""
self.dof_torque = fc.compute_dof_torque(
cfg_ctrl=self.cfg_ctrl,
dof_pos=self.dof_pos,
dof_vel=self.dof_vel,
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
fingertip_midpoint_linvel=self.fingertip_midpoint_linvel,
fingertip_midpoint_angvel=self.fingertip_midpoint_angvel,
left_finger_force=self.left_finger_force,
right_finger_force=self.right_finger_force,
jacobian=self.fingertip_midpoint_jacobian_tf,
arm_mass_matrix=self.arm_mass_matrix,
ctrl_target_gripper_dof_pos=self.ctrl_target_gripper_dof_pos,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
ctrl_target_fingertip_contact_wrench=self.ctrl_target_fingertip_contact_wrench,
device=self.device,
)
self.frankas.set_joint_efforts(efforts=self.dof_torque)
def enable_gravity(self, gravity_mag):
"""Enable gravity."""
gravity = [0.0, 0.0, -gravity_mag]
self._env._world._physics_sim_view.set_gravity(
carb.Float3(gravity[0], gravity[1], gravity[2])
)
def disable_gravity(self):
"""Disable gravity."""
gravity = [0.0, 0.0, 0.0]
self._env._world._physics_sim_view.set_gravity(
carb.Float3(gravity[0], gravity[1], gravity[2])
)
| 26,838 | Python | 45.921329 | 148 | 0.588419 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_schema_config_task.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: schema for task class configurations.
Used by Hydra. Defines template for task class YAML files. Not enforced.
"""
from __future__ import annotations
from dataclasses import dataclass
@dataclass
class Sim:
use_gpu_pipeline: bool # use GPU pipeline
dt: float # timestep size
gravity: list[float] # gravity vector
@dataclass
class Env:
numObservations: int # number of observations per env; camel case required by VecTask
numActions: int # number of actions per env; camel case required by VecTask
numEnvs: int # number of envs; camel case required by VecTask
@dataclass
class Randomize:
franka_arm_initial_dof_pos: list[float] # initial Franka arm DOF position (7)
@dataclass
class RL:
pos_action_scale: list[
float
] # scale on pos displacement targets (3), to convert [-1, 1] to +- x m
rot_action_scale: list[
float
] # scale on rot displacement targets (3), to convert [-1, 1] to +- x rad
force_action_scale: list[
float
] # scale on force targets (3), to convert [-1, 1] to +- x N
torque_action_scale: list[
float
] # scale on torque targets (3), to convert [-1, 1] to +- x Nm
clamp_rot: bool # clamp small values of rotation actions to zero
clamp_rot_thresh: float # smallest acceptable value
max_episode_length: int # max number of timesteps in each episode
@dataclass
class All:
jacobian_type: str # map between joint space and task space via geometric or analytic Jacobian {geometric, analytic}
gripper_prop_gains: list[
float
] # proportional gains on left and right Franka gripper finger DOF position (2)
gripper_deriv_gains: list[
float
] # derivative gains on left and right Franka gripper finger DOF position (2)
@dataclass
class GymDefault:
joint_prop_gains: list[int] # proportional gains on Franka arm DOF position (7)
joint_deriv_gains: list[int] # derivative gains on Franka arm DOF position (7)
@dataclass
class JointSpaceIK:
ik_method: str # use Jacobian pseudoinverse, Jacobian transpose, damped least squares or adaptive SVD {pinv, trans, dls, svd}
joint_prop_gains: list[int]
joint_deriv_gains: list[int]
@dataclass
class JointSpaceID:
ik_method: str
joint_prop_gains: list[int]
joint_deriv_gains: list[int]
@dataclass
class TaskSpaceImpedance:
motion_ctrl_axes: list[bool] # axes for which to enable motion control {0, 1} (6)
task_prop_gains: list[float] # proportional gains on Franka fingertip pose (6)
task_deriv_gains: list[float] # derivative gains on Franka fingertip pose (6)
@dataclass
class OperationalSpaceMotion:
motion_ctrl_axes: list[bool]
task_prop_gains: list[float]
task_deriv_gains: list[float]
@dataclass
class OpenLoopForce:
force_ctrl_axes: list[bool] # axes for which to enable force control {0, 1} (6)
@dataclass
class ClosedLoopForce:
force_ctrl_axes: list[bool]
wrench_prop_gains: list[float] # proportional gains on Franka finger force (6)
@dataclass
class HybridForceMotion:
motion_ctrl_axes: list[bool]
task_prop_gains: list[float]
task_deriv_gains: list[float]
force_ctrl_axes: list[bool]
wrench_prop_gains: list[float]
@dataclass
class Ctrl:
ctrl_type: str # {gym_default,
# joint_space_ik,
# joint_space_id,
# task_space_impedance,
# operational_space_motion,
# open_loop_force,
# closed_loop_force,
# hybrid_force_motion}
gym_default: GymDefault
joint_space_ik: JointSpaceIK
joint_space_id: JointSpaceID
task_space_impedance: TaskSpaceImpedance
operational_space_motion: OperationalSpaceMotion
open_loop_force: OpenLoopForce
closed_loop_force: ClosedLoopForce
hybrid_force_motion: HybridForceMotion
@dataclass
class FactorySchemaConfigTask:
name: str
physics_engine: str
sim: Sim
env: Env
rl: RL
ctrl: Ctrl
| 5,517 | Python | 30.895954 | 130 | 0.719413 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_task_nut_bolt_place.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: Class for nut-bolt place task.
Inherits nut-bolt environment class and abstract task class (not enforced). Can be executed with
PYTHON_PATH omniisaacgymenvs/scripts/rlgames_train.py task=FactoryTaskNutBoltPlace
"""
import asyncio
import hydra
import math
import omegaconf
import torch
from typing import Tuple
import omni.kit
from omni.isaac.core.simulation_context import SimulationContext
import omni.isaac.core.utils.torch as torch_utils
from omni.isaac.core.utils.torch.transformations import tf_combine
import omniisaacgymenvs.tasks.factory.factory_control as fc
from omniisaacgymenvs.tasks.factory.factory_env_nut_bolt import FactoryEnvNutBolt
from omniisaacgymenvs.tasks.factory.factory_schema_class_task import FactoryABCTask
from omniisaacgymenvs.tasks.factory.factory_schema_config_task import (
FactorySchemaConfigTask,
)
class FactoryTaskNutBoltPlace(FactoryEnvNutBolt, FactoryABCTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
"""Initialize environment superclass. Initialize instance variables."""
super().__init__(name, sim_config, env)
self._get_task_yaml_params()
def _get_task_yaml_params(self) -> None:
"""Initialize instance variables from YAML files."""
cs = hydra.core.config_store.ConfigStore.instance()
cs.store(name="factory_schema_config_task", node=FactorySchemaConfigTask)
self.cfg_task = omegaconf.OmegaConf.create(self._task_cfg)
self.max_episode_length = (
self.cfg_task.rl.max_episode_length
) # required instance var for VecTask
asset_info_path = "../tasks/factory/yaml/factory_asset_info_nut_bolt.yaml" # relative to Gym's Hydra search path (cfg dir)
self.asset_info_nut_bolt = hydra.compose(config_name=asset_info_path)
self.asset_info_nut_bolt = self.asset_info_nut_bolt[""][""][""]["tasks"][
"factory"
][
"yaml"
] # strip superfluous nesting
ppo_path = "train/FactoryTaskNutBoltPlacePPO.yaml" # relative to Gym's Hydra search path (cfg dir)
self.cfg_ppo = hydra.compose(config_name=ppo_path)
self.cfg_ppo = self.cfg_ppo["train"] # strip superfluous nesting
def post_reset(self) -> None:
"""Reset the world. Called only once, before simulation begins."""
if self.cfg_task.sim.disable_gravity:
self.disable_gravity()
self.acquire_base_tensors()
self._acquire_task_tensors()
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
# Reset all envs
indices = torch.arange(self.num_envs, dtype=torch.int64, device=self.device)
asyncio.ensure_future(
self.reset_idx_async(indices, randomize_gripper_pose=False)
)
def _acquire_task_tensors(self) -> None:
"""Acquire tensors."""
# Nut-bolt tensors
self.nut_base_pos_local = self.bolt_head_heights * torch.tensor(
[0.0, 0.0, 1.0], device=self.device
).repeat((self.num_envs, 1))
bolt_heights = self.bolt_head_heights + self.bolt_shank_lengths
self.bolt_tip_pos_local = bolt_heights * torch.tensor(
[0.0, 0.0, 1.0], device=self.device
).repeat((self.num_envs, 1))
# Keypoint tensors
self.keypoint_offsets = (
self._get_keypoint_offsets(self.cfg_task.rl.num_keypoints)
* self.cfg_task.rl.keypoint_scale
)
self.keypoints_nut = torch.zeros(
(self.num_envs, self.cfg_task.rl.num_keypoints, 3),
dtype=torch.float32,
device=self.device,
)
self.keypoints_bolt = torch.zeros_like(self.keypoints_nut, device=self.device)
self.identity_quat = (
torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
self.actions = torch.zeros(
(self.num_envs, self.num_actions), device=self.device
)
def pre_physics_step(self, actions) -> None:
"""Reset environments. Apply actions from policy. Simulation step called after this method."""
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(env_ids) > 0:
self.reset_idx(env_ids, randomize_gripper_pose=True)
self.actions = actions.clone().to(
self.device
) # shape = (num_envs, num_actions); values = [-1, 1]
self._apply_actions_as_ctrl_targets(
actions=self.actions, ctrl_target_gripper_dof_pos=0.0, do_scale=True
)
async def pre_physics_step_async(self, actions) -> None:
"""Reset environments. Apply actions from policy. Simulation step called after this method."""
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(env_ids) > 0:
await self.reset_idx_async(env_ids, randomize_gripper_pose=True)
self.actions = actions.clone().to(
self.device
) # shape = (num_envs, num_actions); values = [-1, 1]
self._apply_actions_as_ctrl_targets(
actions=self.actions,
ctrl_target_gripper_dof_pos=0.0,
do_scale=True,
)
def reset_idx(self, env_ids, randomize_gripper_pose) -> None:
"""Reset specified environments."""
self._reset_franka(env_ids)
self._reset_object(env_ids)
# Close gripper onto nut
self.disable_gravity() # to prevent nut from falling
self._close_gripper(sim_steps=self.cfg_task.env.num_gripper_close_sim_steps)
self.enable_gravity(gravity_mag=self.cfg_task.sim.gravity_mag)
if randomize_gripper_pose:
self._randomize_gripper_pose(
env_ids, sim_steps=self.cfg_task.env.num_gripper_move_sim_steps
)
self._reset_buffers(env_ids)
async def reset_idx_async(self, env_ids, randomize_gripper_pose) -> None:
"""Reset specified environments."""
self._reset_franka(env_ids)
self._reset_object(env_ids)
# Close gripper onto nut
self.disable_gravity() # to prevent nut from falling
await self._close_gripper_async(
sim_steps=self.cfg_task.env.num_gripper_close_sim_steps
)
self.enable_gravity(gravity_mag=self.cfg_task.sim.gravity_mag)
if randomize_gripper_pose:
await self._randomize_gripper_pose_async(
env_ids, sim_steps=self.cfg_task.env.num_gripper_move_sim_steps
)
self._reset_buffers(env_ids)
def _reset_franka(self, env_ids) -> None:
"""Reset DOF states and DOF targets of Franka."""
self.dof_pos[env_ids] = torch.cat(
(
torch.tensor(
self.cfg_task.randomize.franka_arm_initial_dof_pos,
device=self.device,
).repeat((len(env_ids), 1)),
(self.nut_widths_max * 0.5)
* 1.1, # buffer on gripper DOF pos to prevent initial contact
(self.nut_widths_max * 0.5) * 1.1,
), # buffer on gripper DOF pos to prevent initial contact
dim=-1,
) # shape = (num_envs, num_dofs)
self.dof_vel[env_ids] = 0.0 # shape = (num_envs, num_dofs)
self.ctrl_target_dof_pos[env_ids] = self.dof_pos[env_ids]
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_positions(self.dof_pos[env_ids], indices=indices)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
def _reset_object(self, env_ids) -> None:
"""Reset root states of nut and bolt."""
# Randomize root state of nut within gripper
self.nut_pos[env_ids, 0] = 0.0
self.nut_pos[env_ids, 1] = 0.0
fingertip_midpoint_pos_reset = 0.58781 # self.fingertip_midpoint_pos at reset
nut_base_pos_local = self.bolt_head_heights.squeeze(-1)
self.nut_pos[env_ids, 2] = fingertip_midpoint_pos_reset - nut_base_pos_local
nut_noise_pos_in_gripper = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
nut_noise_pos_in_gripper = nut_noise_pos_in_gripper @ torch.diag(
torch.tensor(
self.cfg_task.randomize.nut_noise_pos_in_gripper, device=self.device
)
)
self.nut_pos[env_ids, :] += nut_noise_pos_in_gripper[env_ids]
nut_rot_euler = torch.tensor(
[0.0, 0.0, math.pi * 0.5], device=self.device
).repeat(len(env_ids), 1)
nut_noise_rot_in_gripper = 2 * (
torch.rand(self.num_envs, dtype=torch.float32, device=self.device) - 0.5
) # [-1, 1]
nut_noise_rot_in_gripper *= self.cfg_task.randomize.nut_noise_rot_in_gripper
nut_rot_euler[:, 2] += nut_noise_rot_in_gripper
nut_rot_quat = torch_utils.quat_from_euler_xyz(
nut_rot_euler[:, 0], nut_rot_euler[:, 1], nut_rot_euler[:, 2]
)
self.nut_quat[env_ids, :] = nut_rot_quat
self.nut_linvel[env_ids, :] = 0.0
self.nut_angvel[env_ids, :] = 0.0
indices = env_ids.to(dtype=torch.int32)
self.nuts.set_world_poses(
self.nut_pos[env_ids] + self.env_pos[env_ids],
self.nut_quat[env_ids],
indices,
)
self.nuts.set_velocities(
torch.cat((self.nut_linvel[env_ids], self.nut_angvel[env_ids]), dim=1),
indices,
)
# Randomize root state of bolt
bolt_noise_xy = 2 * (
torch.rand((self.num_envs, 2), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
bolt_noise_xy = bolt_noise_xy @ torch.diag(
torch.tensor(
self.cfg_task.randomize.bolt_pos_xy_noise,
dtype=torch.float32,
device=self.device,
)
)
self.bolt_pos[env_ids, 0] = (
self.cfg_task.randomize.bolt_pos_xy_initial[0] + bolt_noise_xy[env_ids, 0]
)
self.bolt_pos[env_ids, 1] = (
self.cfg_task.randomize.bolt_pos_xy_initial[1] + bolt_noise_xy[env_ids, 1]
)
self.bolt_pos[env_ids, 2] = self.cfg_base.env.table_height
self.bolt_quat[env_ids, :] = torch.tensor(
[1.0, 0.0, 0.0, 0.0], dtype=torch.float32, device=self.device
).repeat(len(env_ids), 1)
indices = env_ids.to(dtype=torch.int32)
self.bolts.set_world_poses(
self.bolt_pos[env_ids] + self.env_pos[env_ids],
self.bolt_quat[env_ids],
indices,
)
def _reset_buffers(self, env_ids) -> None:
"""Reset buffers."""
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def _apply_actions_as_ctrl_targets(
self, actions, ctrl_target_gripper_dof_pos, do_scale
) -> None:
"""Apply actions from policy as position/rotation/force/torque targets."""
# Interpret actions as target pos displacements and set pos target
pos_actions = actions[:, 0:3]
if do_scale:
pos_actions = pos_actions @ torch.diag(
torch.tensor(self.cfg_task.rl.pos_action_scale, device=self.device)
)
self.ctrl_target_fingertip_midpoint_pos = (
self.fingertip_midpoint_pos + pos_actions
)
# Interpret actions as target rot (axis-angle) displacements
rot_actions = actions[:, 3:6]
if do_scale:
rot_actions = rot_actions @ torch.diag(
torch.tensor(self.cfg_task.rl.rot_action_scale, device=self.device)
)
# Convert to quat and set rot target
angle = torch.norm(rot_actions, p=2, dim=-1)
axis = rot_actions / angle.unsqueeze(-1)
rot_actions_quat = torch_utils.quat_from_angle_axis(angle, axis)
if self.cfg_task.rl.clamp_rot:
rot_actions_quat = torch.where(
angle.unsqueeze(-1).repeat(1, 4) > self.cfg_task.rl.clamp_rot_thresh,
rot_actions_quat,
torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device).repeat(
self.num_envs, 1
),
)
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_mul(
rot_actions_quat, self.fingertip_midpoint_quat
)
if self.cfg_ctrl["do_force_ctrl"]:
# Interpret actions as target forces and target torques
force_actions = actions[:, 6:9]
if do_scale:
force_actions = force_actions @ torch.diag(
torch.tensor(
self.cfg_task.rl.force_action_scale, device=self.device
)
)
torque_actions = actions[:, 9:12]
if do_scale:
torque_actions = torque_actions @ torch.diag(
torch.tensor(
self.cfg_task.rl.torque_action_scale, device=self.device
)
)
self.ctrl_target_fingertip_contact_wrench = torch.cat(
(force_actions, torque_actions), dim=-1
)
self.ctrl_target_gripper_dof_pos = ctrl_target_gripper_dof_pos
self.generate_ctrl_signals()
def post_physics_step(
self,
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
"""Step buffers. Refresh tensors. Compute observations and reward. Reset environments."""
self.progress_buf[:] += 1
if self._env._world.is_playing():
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
self.get_observations()
self.calculate_metrics()
self.get_extras()
return self.obs_buf, self.rew_buf, self.reset_buf, self.extras
def _refresh_task_tensors(self) -> None:
"""Refresh tensors."""
# Compute pos of keypoints on gripper, nut, and bolt in world frame
for idx, keypoint_offset in enumerate(self.keypoint_offsets):
self.keypoints_nut[:, idx] = tf_combine(
self.nut_quat,
self.nut_pos,
self.identity_quat,
(keypoint_offset + self.nut_base_pos_local),
)[1]
self.keypoints_bolt[:, idx] = tf_combine(
self.bolt_quat,
self.bolt_pos,
self.identity_quat,
(keypoint_offset + self.bolt_tip_pos_local),
)[1]
def get_observations(self) -> dict:
"""Compute observations."""
# Shallow copies of tensors
obs_tensors = [
self.fingertip_midpoint_pos,
self.fingertip_midpoint_quat,
self.fingertip_midpoint_linvel,
self.fingertip_midpoint_angvel,
self.nut_pos,
self.nut_quat,
self.bolt_pos,
self.bolt_quat,
]
if self.cfg_task.rl.add_obs_bolt_tip_pos:
obs_tensors += [self.bolt_tip_pos_local]
self.obs_buf = torch.cat(
obs_tensors, dim=-1
) # shape = (num_envs, num_observations)
observations = {self.frankas.name: {"obs_buf": self.obs_buf}}
return observations
def calculate_metrics(self) -> None:
"""Update reset and reward buffers."""
self._update_reset_buf()
self._update_rew_buf()
def _update_reset_buf(self) -> None:
"""Assign environments for reset if successful or failed."""
# If max episode length has been reached
self.reset_buf[:] = torch.where(
self.progress_buf[:] >= self.max_episode_length - 1,
torch.ones_like(self.reset_buf),
self.reset_buf,
)
def _update_rew_buf(self) -> None:
"""Compute reward at current timestep."""
keypoint_reward = -self._get_keypoint_dist()
action_penalty = (
torch.norm(self.actions, p=2, dim=-1)
* self.cfg_task.rl.action_penalty_scale
)
self.rew_buf[:] = (
keypoint_reward * self.cfg_task.rl.keypoint_reward_scale
- action_penalty * self.cfg_task.rl.action_penalty_scale
)
# In this policy, episode length is constant across all envs
is_last_step = self.progress_buf[0] == self.max_episode_length - 1
if is_last_step:
# Check if nut is close enough to bolt
is_nut_close_to_bolt = self._check_nut_close_to_bolt()
self.rew_buf[:] += is_nut_close_to_bolt * self.cfg_task.rl.success_bonus
self.extras["successes"] = torch.mean(is_nut_close_to_bolt.float())
def _get_keypoint_offsets(self, num_keypoints) -> torch.Tensor:
"""Get uniformly-spaced keypoints along a line of unit length, centered at 0."""
keypoint_offsets = torch.zeros((num_keypoints, 3), device=self.device)
keypoint_offsets[:, -1] = (
torch.linspace(0.0, 1.0, num_keypoints, device=self.device) - 0.5
)
return keypoint_offsets
def _get_keypoint_dist(self) -> torch.Tensor:
"""Get keypoint distance between nut and bolt."""
keypoint_dist = torch.sum(
torch.norm(self.keypoints_bolt - self.keypoints_nut, p=2, dim=-1), dim=-1
)
return keypoint_dist
def _randomize_gripper_pose(self, env_ids, sim_steps) -> None:
"""Move gripper to random pose."""
# Step once to update PhysX with new joint positions and velocities from reset_franka()
SimulationContext.step(self._env._world, render=True)
# Set target pos above table
self.ctrl_target_fingertip_midpoint_pos = torch.tensor(
[0.0, 0.0, self.cfg_base.env.table_height], device=self.device
) + torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_initial, device=self.device
)
self.ctrl_target_fingertip_midpoint_pos = (
self.ctrl_target_fingertip_midpoint_pos.unsqueeze(0).repeat(
self.num_envs, 1
)
)
fingertip_midpoint_pos_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_pos_noise = fingertip_midpoint_pos_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_noise, device=self.device
)
)
self.ctrl_target_fingertip_midpoint_pos += fingertip_midpoint_pos_noise
# Set target rot
ctrl_target_fingertip_midpoint_euler = (
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_initial,
device=self.device,
)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
fingertip_midpoint_rot_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_rot_noise = fingertip_midpoint_rot_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_noise, device=self.device
)
)
ctrl_target_fingertip_midpoint_euler += fingertip_midpoint_rot_noise
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_from_euler_xyz(
ctrl_target_fingertip_midpoint_euler[:, 0],
ctrl_target_fingertip_midpoint_euler[:, 1],
ctrl_target_fingertip_midpoint_euler[:, 2],
)
# Step sim and render
for _ in range(sim_steps):
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
pos_error, axis_angle_error = fc.get_pose_error(
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
jacobian_type=self.cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_hand_pose = torch.cat((pos_error, axis_angle_error), dim=-1)
actions = torch.zeros(
(self.num_envs, self.cfg_task.env.numActions), device=self.device
)
actions[:, :6] = delta_hand_pose
self._apply_actions_as_ctrl_targets(
actions=actions,
ctrl_target_gripper_dof_pos=0.0,
do_scale=False,
)
SimulationContext.step(self._env._world, render=True)
self.dof_vel[env_ids, :] = torch.zeros_like(self.dof_vel[env_ids])
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
# Step once to update PhysX with new joint velocities
SimulationContext.step(self._env._world, render=True)
async def _randomize_gripper_pose_async(self, env_ids, sim_steps) -> None:
"""Move gripper to random pose."""
# Step once to update PhysX with new joint positions and velocities from reset_franka()
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
# Set target pos above table
self.ctrl_target_fingertip_midpoint_pos = torch.tensor(
[0.0, 0.0, self.cfg_base.env.table_height], device=self.device
) + torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_initial, device=self.device
)
self.ctrl_target_fingertip_midpoint_pos = (
self.ctrl_target_fingertip_midpoint_pos.unsqueeze(0).repeat(
self.num_envs, 1
)
)
fingertip_midpoint_pos_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_pos_noise = fingertip_midpoint_pos_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_noise, device=self.device
)
)
self.ctrl_target_fingertip_midpoint_pos += fingertip_midpoint_pos_noise
# Set target rot
ctrl_target_fingertip_midpoint_euler = (
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_initial,
device=self.device,
)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
fingertip_midpoint_rot_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_rot_noise = fingertip_midpoint_rot_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_noise, device=self.device
)
)
ctrl_target_fingertip_midpoint_euler += fingertip_midpoint_rot_noise
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_from_euler_xyz(
ctrl_target_fingertip_midpoint_euler[:, 0],
ctrl_target_fingertip_midpoint_euler[:, 1],
ctrl_target_fingertip_midpoint_euler[:, 2],
)
# Step sim and render
for _ in range(sim_steps):
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
pos_error, axis_angle_error = fc.get_pose_error(
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
jacobian_type=self.cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_hand_pose = torch.cat((pos_error, axis_angle_error), dim=-1)
actions = torch.zeros(
(self.num_envs, self.cfg_task.env.numActions), device=self.device
)
actions[:, :6] = delta_hand_pose
self._apply_actions_as_ctrl_targets(
actions=actions,
ctrl_target_gripper_dof_pos=0.0,
do_scale=False,
)
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
self.dof_vel[env_ids, :] = torch.zeros_like(self.dof_vel[env_ids])
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
# Step once to update PhysX with new joint velocities
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
def _close_gripper(self, sim_steps) -> None:
"""Fully close gripper using controller. Called outside RL loop (i.e., after last step of episode)."""
self._move_gripper_to_dof_pos(gripper_dof_pos=0.0, sim_steps=sim_steps)
def _move_gripper_to_dof_pos(self, gripper_dof_pos, sim_steps) -> None:
"""Move gripper fingers to specified DOF position using controller."""
delta_hand_pose = torch.zeros(
(self.num_envs, 6), device=self.device
) # No hand motion
# Step sim
for _ in range(sim_steps):
self._apply_actions_as_ctrl_targets(
delta_hand_pose, gripper_dof_pos, do_scale=False
)
SimulationContext.step(self._env._world, render=True)
async def _close_gripper_async(self, sim_steps) -> None:
"""Fully close gripper using controller. Called outside RL loop (i.e., after last step of episode)."""
await self._move_gripper_to_dof_pos_async(
gripper_dof_pos=0.0, sim_steps=sim_steps
)
async def _move_gripper_to_dof_pos_async(
self, gripper_dof_pos, sim_steps
) -> None:
"""Move gripper fingers to specified DOF position using controller."""
delta_hand_pose = torch.zeros(
(self.num_envs, 6), device=self.device
) # No hand motion
# Step sim
for _ in range(sim_steps):
self._apply_actions_as_ctrl_targets(
delta_hand_pose, gripper_dof_pos, do_scale=False
)
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
def _check_nut_close_to_bolt(self) -> torch.Tensor:
"""Check if nut is close to bolt."""
keypoint_dist = torch.norm(
self.keypoints_bolt - self.keypoints_nut, p=2, dim=-1
)
is_nut_close_to_bolt = torch.where(
torch.sum(keypoint_dist, dim=-1) < self.cfg_task.rl.close_error_thresh,
torch.ones_like(self.progress_buf),
torch.zeros_like(self.progress_buf),
)
return is_nut_close_to_bolt
| 29,034 | Python | 37.868809 | 131 | 0.594303 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_schema_config_env.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: schema for environment class configurations.
Used by Hydra. Defines template for environment class YAML files.
"""
from dataclasses import dataclass
@dataclass
class Sim:
disable_franka_collisions: bool # disable collisions between Franka and objects
@dataclass
class Env:
env_name: str # name of scene
@dataclass
class FactorySchemaConfigEnv:
sim: Sim
env: Env
| 1,960 | Python | 36.711538 | 84 | 0.776531 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_schema_class_task.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: abstract base class for task classes.
Inherits ABC class. Inherited by task classes. Defines template for task classes.
"""
from abc import ABC, abstractmethod
class FactoryABCTask(ABC):
@abstractmethod
def __init__(self):
"""Initialize instance variables. Initialize environment superclass."""
pass
@abstractmethod
def _get_task_yaml_params(self):
"""Initialize instance variables from YAML files."""
pass
@abstractmethod
def _acquire_task_tensors(self):
"""Acquire tensors."""
pass
@abstractmethod
def _refresh_task_tensors(self):
"""Refresh tensors."""
pass
@abstractmethod
def pre_physics_step(self):
"""Reset environments. Apply actions from policy as controller targets. Simulation step called after this method."""
pass
@abstractmethod
def post_physics_step(self):
"""Step buffers. Refresh tensors. Compute observations and reward."""
pass
@abstractmethod
def get_observations(self):
"""Compute observations."""
pass
@abstractmethod
def calculate_metrics(self):
"""Detect successes and failures. Update reward and reset buffers."""
pass
@abstractmethod
def _update_rew_buf(self):
"""Compute reward at current timestep."""
pass
@abstractmethod
def _update_reset_buf(self):
"""Assign environments for reset if successful or failed."""
pass
@abstractmethod
def reset_idx(self):
"""Reset specified environments."""
pass
@abstractmethod
def _reset_franka(self):
"""Reset DOF states and DOF targets of Franka."""
pass
@abstractmethod
def _reset_object(self):
"""Reset root state of object."""
pass
@abstractmethod
def _reset_buffers(self):
"""Reset buffers."""
pass
| 3,492 | Python | 31.342592 | 124 | 0.69559 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_schema_class_env.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: abstract base class for environment classes.
Inherits ABC class. Inherited by environment classes. Defines template for environment classes.
"""
from abc import ABC, abstractmethod
class FactoryABCEnv(ABC):
@abstractmethod
def __init__(self):
"""Initialize instance variables. Initialize base superclass. Acquire tensors."""
pass
@abstractmethod
def _get_env_yaml_params(self):
"""Initialize instance variables from YAML files."""
pass
@abstractmethod
def set_up_scene(self):
"""Set env options. Import assets. Create actors."""
pass
@abstractmethod
def _import_env_assets(self):
"""Set asset options. Import assets."""
pass
@abstractmethod
def refresh_env_tensors(self):
"""Refresh tensors."""
# NOTE: Tensor refresh functions should be called once per step, before setters.
pass
| 2,489 | Python | 37.906249 | 95 | 0.73644 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_task_nut_bolt_screw.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: Class for nut-bolt screw task.
Inherits nut-bolt environment class and abstract task class (not enforced). Can be executed with
PYTHON_PATH omniisaacgymenvs/scripts/rlgames_train.py task=FactoryTaskNutBoltScrew
"""
import hydra
import math
import omegaconf
import torch
from typing import Tuple
import omni.isaac.core.utils.torch as torch_utils
import omniisaacgymenvs.tasks.factory.factory_control as fc
from omniisaacgymenvs.tasks.factory.factory_env_nut_bolt import FactoryEnvNutBolt
from omniisaacgymenvs.tasks.factory.factory_schema_class_task import FactoryABCTask
from omniisaacgymenvs.tasks.factory.factory_schema_config_task import (
FactorySchemaConfigTask,
)
class FactoryTaskNutBoltScrew(FactoryEnvNutBolt, FactoryABCTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
"""Initialize environment superclass. Initialize instance variables."""
super().__init__(name, sim_config, env)
self._get_task_yaml_params()
def _get_task_yaml_params(self) -> None:
"""Initialize instance variables from YAML files."""
cs = hydra.core.config_store.ConfigStore.instance()
cs.store(name="factory_schema_config_task", node=FactorySchemaConfigTask)
self.cfg_task = omegaconf.OmegaConf.create(self._task_cfg)
self.max_episode_length = (
self.cfg_task.rl.max_episode_length
) # required instance var for VecTask
asset_info_path = "../tasks/factory/yaml/factory_asset_info_nut_bolt.yaml" # relative to Gym's Hydra search path (cfg dir)
self.asset_info_nut_bolt = hydra.compose(config_name=asset_info_path)
self.asset_info_nut_bolt = self.asset_info_nut_bolt[""][""][""]["tasks"][
"factory"
][
"yaml"
] # strip superfluous nesting
ppo_path = "train/FactoryTaskNutBoltScrewPPO.yaml" # relative to Gym's Hydra search path (cfg dir)
self.cfg_ppo = hydra.compose(config_name=ppo_path)
self.cfg_ppo = self.cfg_ppo["train"] # strip superfluous nesting
def post_reset(self) -> None:
"""Reset the world. Called only once, before simulation begins."""
if self.cfg_task.sim.disable_gravity:
self.disable_gravity()
self.acquire_base_tensors()
self._acquire_task_tensors()
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
# Reset all envs
indices = torch.arange(self.num_envs, dtype=torch.int64, device=self.device)
self.reset_idx(indices)
def _acquire_task_tensors(self) -> None:
"""Acquire tensors."""
target_heights = (
self.cfg_base.env.table_height
+ self.bolt_head_heights
+ self.nut_heights * 0.5
)
self.target_pos = target_heights * torch.tensor(
[0.0, 0.0, 1.0], device=self.device
).repeat((self.num_envs, 1))
self.identity_quat = (
torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
self.actions = torch.zeros(
(self.num_envs, self.num_actions), device=self.device
)
def pre_physics_step(self, actions) -> None:
"""Reset environments. Apply actions from policy. Simulation step called after this method."""
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(env_ids) > 0:
self.reset_idx(env_ids)
self.actions = actions.clone().to(
self.device
) # shape = (num_envs, num_actions); values = [-1, 1]
self._apply_actions_as_ctrl_targets(
actions=self.actions, ctrl_target_gripper_dof_pos=0.0, do_scale=True
)
def reset_idx(self, env_ids) -> None:
"""Reset specified environments."""
self._reset_franka(env_ids)
self._reset_object(env_ids)
self._reset_buffers(env_ids)
def _reset_franka(self, env_ids) -> None:
"""Reset DOF states and DOF targets of Franka."""
self.dof_pos[env_ids] = torch.cat(
(
torch.tensor(
self.cfg_task.randomize.franka_arm_initial_dof_pos,
device=self.device,
).repeat((len(env_ids), 1)),
(self.nut_widths_max[env_ids] * 0.5)
* 1.1, # buffer on gripper DOF pos to prevent initial contact
(self.nut_widths_max[env_ids] * 0.5) * 1.1,
), # buffer on gripper DOF pos to prevent initial contact
dim=-1,
) # shape = (num_envs, num_dofs)
self.dof_vel[env_ids] = 0.0 # shape = (num_envs, num_dofs)
self.ctrl_target_dof_pos[env_ids] = self.dof_pos[env_ids]
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_positions(self.dof_pos[env_ids], indices=indices)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
def _reset_object(self, env_ids) -> None:
"""Reset root state of nut."""
nut_pos = self.cfg_base.env.table_height + self.bolt_shank_lengths[env_ids]
self.nut_pos[env_ids, :] = nut_pos * torch.tensor(
[0.0, 0.0, 1.0], device=self.device
).repeat(len(env_ids), 1)
nut_rot = (
self.cfg_task.randomize.nut_rot_initial
* torch.ones((len(env_ids), 1), device=self.device)
* math.pi
/ 180.0
)
self.nut_quat[env_ids, :] = torch.cat(
(
torch.cos(nut_rot * 0.5),
torch.zeros((len(env_ids), 1), device=self.device),
torch.zeros((len(env_ids), 1), device=self.device),
torch.sin(nut_rot * 0.5),
),
dim=-1,
)
self.nut_linvel[env_ids, :] = 0.0
self.nut_angvel[env_ids, :] = 0.0
indices = env_ids.to(dtype=torch.int32)
self.nuts.set_world_poses(
self.nut_pos[env_ids] + self.env_pos[env_ids],
self.nut_quat[env_ids],
indices,
)
self.nuts.set_velocities(
torch.cat((self.nut_linvel[env_ids], self.nut_angvel[env_ids]), dim=1),
indices,
)
def _reset_buffers(self, env_ids) -> None:
"""Reset buffers."""
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def _apply_actions_as_ctrl_targets(
self, actions, ctrl_target_gripper_dof_pos, do_scale
) -> None:
"""Apply actions from policy as position/rotation/force/torque targets."""
# Interpret actions as target pos displacements and set pos target
pos_actions = actions[:, 0:3]
if self.cfg_task.rl.unidirectional_pos:
pos_actions[:, 2] = -(pos_actions[:, 2] + 1.0) * 0.5 # [-1, 0]
if do_scale:
pos_actions = pos_actions @ torch.diag(
torch.tensor(self.cfg_task.rl.pos_action_scale, device=self.device)
)
self.ctrl_target_fingertip_midpoint_pos = (
self.fingertip_midpoint_pos + pos_actions
)
# Interpret actions as target rot (axis-angle) displacements
rot_actions = actions[:, 3:6]
if self.cfg_task.rl.unidirectional_rot:
rot_actions[:, 2] = -(rot_actions[:, 2] + 1.0) * 0.5 # [-1, 0]
if do_scale:
rot_actions = rot_actions @ torch.diag(
torch.tensor(self.cfg_task.rl.rot_action_scale, device=self.device)
)
# Convert to quat and set rot target
angle = torch.norm(rot_actions, p=2, dim=-1)
axis = rot_actions / angle.unsqueeze(-1)
rot_actions_quat = torch_utils.quat_from_angle_axis(angle, axis)
if self.cfg_task.rl.clamp_rot:
rot_actions_quat = torch.where(
angle.unsqueeze(-1).repeat(1, 4) > self.cfg_task.rl.clamp_rot_thresh,
rot_actions_quat,
torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device).repeat(
self.num_envs, 1
),
)
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_mul(
rot_actions_quat, self.fingertip_midpoint_quat
)
if self.cfg_ctrl["do_force_ctrl"]:
# Interpret actions as target forces and target torques
force_actions = actions[:, 6:9]
if self.cfg_task.rl.unidirectional_force:
force_actions[:, 2] = -(force_actions[:, 2] + 1.0) * 0.5 # [-1, 0]
if do_scale:
force_actions = force_actions @ torch.diag(
torch.tensor(
self.cfg_task.rl.force_action_scale, device=self.device
)
)
torque_actions = actions[:, 9:12]
if do_scale:
torque_actions = torque_actions @ torch.diag(
torch.tensor(
self.cfg_task.rl.torque_action_scale, device=self.device
)
)
self.ctrl_target_fingertip_contact_wrench = torch.cat(
(force_actions, torque_actions), dim=-1
)
self.ctrl_target_gripper_dof_pos = ctrl_target_gripper_dof_pos
self.generate_ctrl_signals()
def post_physics_step(
self,
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
"""Step buffers. Refresh tensors. Compute observations and reward. Reset environments."""
self.progress_buf[:] += 1
if self._env._world.is_playing():
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
self.get_observations()
self.calculate_metrics()
self.get_extras()
return self.obs_buf, self.rew_buf, self.reset_buf, self.extras
def _refresh_task_tensors(self) -> None:
"""Refresh tensors."""
self.fingerpad_midpoint_pos = fc.translate_along_local_z(
pos=self.finger_midpoint_pos,
quat=self.hand_quat,
offset=self.asset_info_franka_table.franka_finger_length
- self.asset_info_franka_table.franka_fingerpad_length * 0.5,
device=self.device,
)
self.finger_nut_keypoint_dist = self._get_keypoint_dist(body="finger_nut")
self.nut_keypoint_dist = self._get_keypoint_dist(body="nut")
self.nut_dist_to_target = torch.norm(
self.target_pos - self.nut_com_pos, p=2, dim=-1
) # distance between nut COM and target
self.nut_dist_to_fingerpads = torch.norm(
self.fingerpad_midpoint_pos - self.nut_com_pos, p=2, dim=-1
) # distance between nut COM and midpoint between centers of fingerpads
self.was_success = torch.zeros_like(self.progress_buf, dtype=torch.bool)
def get_observations(self) -> dict:
"""Compute observations."""
# Shallow copies of tensors
obs_tensors = [
self.fingertip_midpoint_pos,
self.fingertip_midpoint_quat,
self.fingertip_midpoint_linvel,
self.fingertip_midpoint_angvel,
self.nut_com_pos,
self.nut_com_quat,
self.nut_com_linvel,
self.nut_com_angvel,
]
if self.cfg_task.rl.add_obs_finger_force:
obs_tensors += [self.left_finger_force, self.right_finger_force]
else:
obs_tensors += [
torch.zeros_like(self.left_finger_force),
torch.zeros_like(self.right_finger_force),
]
self.obs_buf = torch.cat(
obs_tensors, dim=-1
) # shape = (num_envs, num_observations)
observations = {self.frankas.name: {"obs_buf": self.obs_buf}}
return observations
def calculate_metrics(self) -> None:
"""Update reset and reward buffers."""
# Get successful and failed envs at current timestep
curr_successes = self._get_curr_successes()
curr_failures = self._get_curr_failures(curr_successes)
self._update_reset_buf(curr_successes, curr_failures)
self._update_rew_buf(curr_successes)
if torch.any(self.is_expired):
self.extras["successes"] = torch.mean(curr_successes.float())
def _update_reset_buf(self, curr_successes, curr_failures) -> None:
"""Assign environments for reset if successful or failed."""
self.reset_buf[:] = self.is_expired
def _update_rew_buf(self, curr_successes) -> None:
"""Compute reward at current timestep."""
keypoint_reward = -(self.nut_keypoint_dist + self.finger_nut_keypoint_dist)
action_penalty = torch.norm(self.actions, p=2, dim=-1)
self.rew_buf[:] = (
keypoint_reward * self.cfg_task.rl.keypoint_reward_scale
- action_penalty * self.cfg_task.rl.action_penalty_scale
+ curr_successes * self.cfg_task.rl.success_bonus
)
def _get_keypoint_dist(self, body) -> torch.Tensor:
"""Get keypoint distance."""
axis_length = (
self.asset_info_franka_table.franka_hand_length
+ self.asset_info_franka_table.franka_finger_length
)
if body == "finger" or body == "nut":
# Keypoint distance between finger/nut and target
if body == "finger":
self.keypoint1 = self.fingertip_midpoint_pos
self.keypoint2 = fc.translate_along_local_z(
pos=self.keypoint1,
quat=self.fingertip_midpoint_quat,
offset=-axis_length,
device=self.device,
)
elif body == "nut":
self.keypoint1 = self.nut_com_pos
self.keypoint2 = fc.translate_along_local_z(
pos=self.nut_com_pos,
quat=self.nut_com_quat,
offset=axis_length,
device=self.device,
)
self.keypoint1_targ = self.target_pos
self.keypoint2_targ = self.keypoint1_targ + torch.tensor(
[0.0, 0.0, axis_length], device=self.device
)
elif body == "finger_nut":
# Keypoint distance between finger and nut
self.keypoint1 = self.fingerpad_midpoint_pos
self.keypoint2 = fc.translate_along_local_z(
pos=self.keypoint1,
quat=self.fingertip_midpoint_quat,
offset=-axis_length,
device=self.device,
)
self.keypoint1_targ = self.nut_com_pos
self.keypoint2_targ = fc.translate_along_local_z(
pos=self.nut_com_pos,
quat=self.nut_com_quat,
offset=axis_length,
device=self.device,
)
self.keypoint3 = self.keypoint1 + (self.keypoint2 - self.keypoint1) * 1.0 / 3.0
self.keypoint4 = self.keypoint1 + (self.keypoint2 - self.keypoint1) * 2.0 / 3.0
self.keypoint3_targ = (
self.keypoint1_targ
+ (self.keypoint2_targ - self.keypoint1_targ) * 1.0 / 3.0
)
self.keypoint4_targ = (
self.keypoint1_targ
+ (self.keypoint2_targ - self.keypoint1_targ) * 2.0 / 3.0
)
keypoint_dist = (
torch.norm(self.keypoint1_targ - self.keypoint1, p=2, dim=-1)
+ torch.norm(self.keypoint2_targ - self.keypoint2, p=2, dim=-1)
+ torch.norm(self.keypoint3_targ - self.keypoint3, p=2, dim=-1)
+ torch.norm(self.keypoint4_targ - self.keypoint4, p=2, dim=-1)
)
return keypoint_dist
def _get_curr_successes(self) -> torch.Tensor:
"""Get success mask at current timestep."""
curr_successes = torch.zeros(
(self.num_envs,), dtype=torch.bool, device=self.device
)
# If nut is close enough to target pos
is_close = torch.where(
self.nut_dist_to_target < self.thread_pitches.squeeze(-1) * 5,
torch.ones_like(curr_successes),
torch.zeros_like(curr_successes),
)
curr_successes = torch.logical_or(curr_successes, is_close)
return curr_successes
def _get_curr_failures(self, curr_successes) -> torch.Tensor:
"""Get failure mask at current timestep."""
curr_failures = torch.zeros(
(self.num_envs,), dtype=torch.bool, device=self.device
)
# If max episode length has been reached
self.is_expired = torch.where(
self.progress_buf[:] >= self.cfg_task.rl.max_episode_length,
torch.ones_like(curr_failures),
curr_failures,
)
# If nut is too far from target pos
self.is_far = torch.where(
self.nut_dist_to_target > self.cfg_task.rl.far_error_thresh,
torch.ones_like(curr_failures),
curr_failures,
)
# If nut has slipped (distance-based definition)
self.is_slipped = torch.where(
self.nut_dist_to_fingerpads
> self.asset_info_franka_table.franka_fingerpad_length * 0.5
+ self.nut_heights.squeeze(-1) * 0.5,
torch.ones_like(curr_failures),
curr_failures,
)
self.is_slipped = torch.logical_and(
self.is_slipped, torch.logical_not(curr_successes)
) # ignore slip if successful
# If nut has fallen (i.e., if nut XY pos has drifted from center of bolt and nut Z pos has drifted below top of bolt)
self.is_fallen = torch.logical_and(
torch.norm(self.nut_com_pos[:, 0:2], p=2, dim=-1)
> self.bolt_widths.squeeze(-1) * 0.5,
self.nut_com_pos[:, 2]
< self.cfg_base.env.table_height
+ self.bolt_head_heights.squeeze(-1)
+ self.bolt_shank_lengths.squeeze(-1)
+ self.nut_heights.squeeze(-1) * 0.5,
)
curr_failures = torch.logical_or(curr_failures, self.is_expired)
curr_failures = torch.logical_or(curr_failures, self.is_far)
curr_failures = torch.logical_or(curr_failures, self.is_slipped)
curr_failures = torch.logical_or(curr_failures, self.is_fallen)
return curr_failures
| 20,039 | Python | 37.390805 | 131 | 0.589051 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_task_nut_bolt_pick.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: Class for nut-bolt pick task.
Inherits nut-bolt environment class and abstract task class (not enforced). Can be executed with
PYTHON_PATH omniisaacgymenvs/scripts/rlgames_train.py task=FactoryTaskNutBoltPick
"""
import asyncio
import hydra
import omegaconf
import torch
import omni.kit
from omni.isaac.core.simulation_context import SimulationContext
from omni.isaac.core.utils.torch.transformations import tf_combine
from typing import Tuple
import omni.isaac.core.utils.torch as torch_utils
import omniisaacgymenvs.tasks.factory.factory_control as fc
from omniisaacgymenvs.tasks.factory.factory_env_nut_bolt import FactoryEnvNutBolt
from omniisaacgymenvs.tasks.factory.factory_schema_class_task import FactoryABCTask
from omniisaacgymenvs.tasks.factory.factory_schema_config_task import (
FactorySchemaConfigTask,
)
class FactoryTaskNutBoltPick(FactoryEnvNutBolt, FactoryABCTask):
def __init__(self, name, sim_config, env, offset=None) -> None:
"""Initialize environment superclass. Initialize instance variables."""
super().__init__(name, sim_config, env)
self._get_task_yaml_params()
def _get_task_yaml_params(self) -> None:
"""Initialize instance variables from YAML files."""
cs = hydra.core.config_store.ConfigStore.instance()
cs.store(name="factory_schema_config_task", node=FactorySchemaConfigTask)
self.cfg_task = omegaconf.OmegaConf.create(self._task_cfg)
self.max_episode_length = (
self.cfg_task.rl.max_episode_length
) # required instance var for VecTask
asset_info_path = "../tasks/factory/yaml/factory_asset_info_nut_bolt.yaml" # relative to Gym's Hydra search path (cfg dir)
self.asset_info_nut_bolt = hydra.compose(config_name=asset_info_path)
self.asset_info_nut_bolt = self.asset_info_nut_bolt[""][""][""]["tasks"][
"factory"
][
"yaml"
] # strip superfluous nesting
ppo_path = "train/FactoryTaskNutBoltPickPPO.yaml" # relative to Gym's Hydra search path (cfg dir)
self.cfg_ppo = hydra.compose(config_name=ppo_path)
self.cfg_ppo = self.cfg_ppo["train"] # strip superfluous nesting
def post_reset(self) -> None:
"""Reset the world. Called only once, before simulation begins."""
if self.cfg_task.sim.disable_gravity:
self.disable_gravity()
self.acquire_base_tensors()
self._acquire_task_tensors()
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
# Reset all envs
indices = torch.arange(self._num_envs, dtype=torch.int64, device=self._device)
asyncio.ensure_future(
self.reset_idx_async(indices, randomize_gripper_pose=False)
)
def _acquire_task_tensors(self) -> None:
"""Acquire tensors."""
# Grasp pose tensors
nut_grasp_heights = self.bolt_head_heights + self.nut_heights * 0.5 # nut COM
self.nut_grasp_pos_local = nut_grasp_heights * torch.tensor(
[0.0, 0.0, 1.0], device=self.device
).repeat((self.num_envs, 1))
self.nut_grasp_quat_local = (
torch.tensor([0.0, 0.0, 1.0, 0.0], device=self.device)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
# Keypoint tensors
self.keypoint_offsets = (
self._get_keypoint_offsets(self.cfg_task.rl.num_keypoints)
* self.cfg_task.rl.keypoint_scale
)
self.keypoints_gripper = torch.zeros(
(self.num_envs, self.cfg_task.rl.num_keypoints, 3),
dtype=torch.float32,
device=self.device,
)
self.keypoints_nut = torch.zeros_like(
self.keypoints_gripper, device=self.device
)
self.identity_quat = (
torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
self.actions = torch.zeros(
(self.num_envs, self.num_actions), device=self.device
)
def pre_physics_step(self, actions) -> None:
"""Reset environments. Apply actions from policy. Simulation step called after this method."""
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(env_ids) > 0:
self.reset_idx(env_ids, randomize_gripper_pose=True)
self.actions = actions.clone().to(
self.device
) # shape = (num_envs, num_actions); values = [-1, 1]
self._apply_actions_as_ctrl_targets(
actions=self.actions,
ctrl_target_gripper_dof_pos=self.asset_info_franka_table.franka_gripper_width_max,
do_scale=True,
)
async def pre_physics_step_async(self, actions) -> None:
"""Reset environments. Apply actions from policy. Simulation step called after this method."""
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(env_ids) > 0:
await self.reset_idx_async(env_ids, randomize_gripper_pose=True)
self.actions = actions.clone().to(
self.device
) # shape = (num_envs, num_actions); values = [-1, 1]
self._apply_actions_as_ctrl_targets(
actions=self.actions,
ctrl_target_gripper_dof_pos=self.asset_info_franka_table.franka_gripper_width_max,
do_scale=True,
)
def reset_idx(self, env_ids, randomize_gripper_pose) -> None:
"""Reset specified environments."""
self._reset_franka(env_ids)
self._reset_object(env_ids)
if randomize_gripper_pose:
self._randomize_gripper_pose(
env_ids, sim_steps=self.cfg_task.env.num_gripper_move_sim_steps
)
self._reset_buffers(env_ids)
async def reset_idx_async(self, env_ids, randomize_gripper_pose) -> None:
"""Reset specified environments."""
self._reset_franka(env_ids)
self._reset_object(env_ids)
if randomize_gripper_pose:
await self._randomize_gripper_pose_async(
env_ids, sim_steps=self.cfg_task.env.num_gripper_move_sim_steps
)
self._reset_buffers(env_ids)
def _reset_franka(self, env_ids) -> None:
"""Reset DOF states and DOF targets of Franka."""
self.dof_pos[env_ids] = torch.cat(
(
torch.tensor(
self.cfg_task.randomize.franka_arm_initial_dof_pos,
device=self.device,
),
torch.tensor(
[self.asset_info_franka_table.franka_gripper_width_max],
device=self.device,
),
torch.tensor(
[self.asset_info_franka_table.franka_gripper_width_max],
device=self.device,
),
),
dim=-1,
) # shape = (num_envs, num_dofs)
self.dof_vel[env_ids] = 0.0 # shape = (num_envs, num_dofs)
self.ctrl_target_dof_pos[env_ids] = self.dof_pos[env_ids]
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_positions(self.dof_pos[env_ids], indices=indices)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
def _reset_object(self, env_ids) -> None:
"""Reset root states of nut and bolt."""
# Randomize root state of nut
nut_noise_xy = 2 * (
torch.rand((self.num_envs, 2), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
nut_noise_xy = nut_noise_xy @ torch.diag(
torch.tensor(self.cfg_task.randomize.nut_pos_xy_noise, device=self.device)
)
self.nut_pos[env_ids, 0] = (
self.cfg_task.randomize.nut_pos_xy_initial[0] + nut_noise_xy[env_ids, 0]
)
self.nut_pos[env_ids, 1] = (
self.cfg_task.randomize.nut_pos_xy_initial[1] + nut_noise_xy[env_ids, 1]
)
self.nut_pos[
env_ids, 2
] = self.cfg_base.env.table_height - self.bolt_head_heights.squeeze(-1)
self.nut_quat[env_ids, :] = torch.tensor(
[1.0, 0.0, 0.0, 0.0], dtype=torch.float32, device=self.device
).repeat(len(env_ids), 1)
self.nut_linvel[env_ids, :] = 0.0
self.nut_angvel[env_ids, :] = 0.0
indices = env_ids.to(dtype=torch.int32)
self.nuts.set_world_poses(
self.nut_pos[env_ids] + self.env_pos[env_ids],
self.nut_quat[env_ids],
indices,
)
self.nuts.set_velocities(
torch.cat((self.nut_linvel[env_ids], self.nut_angvel[env_ids]), dim=1),
indices,
)
# Randomize root state of bolt
bolt_noise_xy = 2 * (
torch.rand((self.num_envs, 2), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
bolt_noise_xy = bolt_noise_xy @ torch.diag(
torch.tensor(self.cfg_task.randomize.bolt_pos_xy_noise, device=self.device)
)
self.bolt_pos[env_ids, 0] = (
self.cfg_task.randomize.bolt_pos_xy_initial[0] + bolt_noise_xy[env_ids, 0]
)
self.bolt_pos[env_ids, 1] = (
self.cfg_task.randomize.bolt_pos_xy_initial[1] + bolt_noise_xy[env_ids, 1]
)
self.bolt_pos[env_ids, 2] = self.cfg_base.env.table_height
self.bolt_quat[env_ids, :] = torch.tensor(
[1.0, 0.0, 0.0, 0.0], dtype=torch.float32, device=self.device
).repeat(len(env_ids), 1)
indices = env_ids.to(dtype=torch.int32)
self.bolts.set_world_poses(
self.bolt_pos[env_ids] + self.env_pos[env_ids],
self.bolt_quat[env_ids],
indices,
)
def _reset_buffers(self, env_ids) -> None:
"""Reset buffers."""
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
def _apply_actions_as_ctrl_targets(
self, actions, ctrl_target_gripper_dof_pos, do_scale
) -> None:
"""Apply actions from policy as position/rotation/force/torque targets."""
# Interpret actions as target pos displacements and set pos target
pos_actions = actions[:, 0:3]
if do_scale:
pos_actions = pos_actions @ torch.diag(
torch.tensor(self.cfg_task.rl.pos_action_scale, device=self.device)
)
self.ctrl_target_fingertip_midpoint_pos = (
self.fingertip_midpoint_pos + pos_actions
)
# Interpret actions as target rot (axis-angle) displacements
rot_actions = actions[:, 3:6]
if do_scale:
rot_actions = rot_actions @ torch.diag(
torch.tensor(self.cfg_task.rl.rot_action_scale, device=self.device)
)
# Convert to quat and set rot target
angle = torch.norm(rot_actions, p=2, dim=-1)
axis = rot_actions / angle.unsqueeze(-1)
rot_actions_quat = torch_utils.quat_from_angle_axis(angle, axis)
if self.cfg_task.rl.clamp_rot:
rot_actions_quat = torch.where(
angle.unsqueeze(-1).repeat(1, 4) > self.cfg_task.rl.clamp_rot_thresh,
rot_actions_quat,
torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device).repeat(
self.num_envs, 1
),
)
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_mul(
rot_actions_quat, self.fingertip_midpoint_quat
)
if self.cfg_ctrl["do_force_ctrl"]:
# Interpret actions as target forces and target torques
force_actions = actions[:, 6:9]
if do_scale:
force_actions = force_actions @ torch.diag(
torch.tensor(
self.cfg_task.rl.force_action_scale, device=self.device
)
)
torque_actions = actions[:, 9:12]
if do_scale:
torque_actions = torque_actions @ torch.diag(
torch.tensor(
self.cfg_task.rl.torque_action_scale, device=self.device
)
)
self.ctrl_target_fingertip_contact_wrench = torch.cat(
(force_actions, torque_actions), dim=-1
)
self.ctrl_target_gripper_dof_pos = ctrl_target_gripper_dof_pos
self.generate_ctrl_signals()
def post_physics_step(
self,
) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor]:
"""Step buffers. Refresh tensors. Compute observations and reward. Reset environments."""
self.progress_buf[:] += 1
if self._env._world.is_playing():
# In this policy, episode length is constant
is_last_step = self.progress_buf[0] == self.max_episode_length - 1
if is_last_step:
# At this point, robot has executed RL policy. Now close gripper and lift (open-loop)
if self.cfg_task.env.close_and_lift:
self._close_gripper(
sim_steps=self.cfg_task.env.num_gripper_close_sim_steps
)
self._lift_gripper(
franka_gripper_width=0.0,
lift_distance=0.3,
sim_steps=self.cfg_task.env.num_gripper_lift_sim_steps,
)
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
self.get_observations()
self.get_states()
self.calculate_metrics()
self.get_extras()
return self.obs_buf, self.rew_buf, self.reset_buf, self.extras
async def post_physics_step_async(self):
"""Step buffers. Refresh tensors. Compute observations and reward. Reset environments."""
self.progress_buf[:] += 1
if self._env._world.is_playing():
# In this policy, episode length is constant
is_last_step = self.progress_buf[0] == self.max_episode_length - 1
if self.cfg_task.env.close_and_lift:
# At this point, robot has executed RL policy. Now close gripper and lift (open-loop)
if is_last_step:
await self._close_gripper_async(
sim_steps=self.cfg_task.env.num_gripper_close_sim_steps
)
await self._lift_gripper_async(
sim_steps=self.cfg_task.env.num_gripper_lift_sim_steps
)
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
self.get_observations()
self.get_states()
self.calculate_metrics()
self.get_extras()
return self.obs_buf, self.rew_buf, self.reset_buf, self.extras
def _refresh_task_tensors(self):
"""Refresh tensors."""
# Compute pose of nut grasping frame
self.nut_grasp_quat, self.nut_grasp_pos = tf_combine(
self.nut_quat,
self.nut_pos,
self.nut_grasp_quat_local,
self.nut_grasp_pos_local,
)
# Compute pos of keypoints on gripper and nut in world frame
for idx, keypoint_offset in enumerate(self.keypoint_offsets):
self.keypoints_gripper[:, idx] = tf_combine(
self.fingertip_midpoint_quat,
self.fingertip_midpoint_pos,
self.identity_quat,
keypoint_offset.repeat(self.num_envs, 1),
)[1]
self.keypoints_nut[:, idx] = tf_combine(
self.nut_grasp_quat,
self.nut_grasp_pos,
self.identity_quat,
keypoint_offset.repeat(self.num_envs, 1),
)[1]
def get_observations(self) -> dict:
"""Compute observations."""
# Shallow copies of tensors
obs_tensors = [
self.fingertip_midpoint_pos,
self.fingertip_midpoint_quat,
self.fingertip_midpoint_linvel,
self.fingertip_midpoint_angvel,
self.nut_grasp_pos,
self.nut_grasp_quat,
]
self.obs_buf = torch.cat(
obs_tensors, dim=-1
) # shape = (num_envs, num_observations)
observations = {self.frankas.name: {"obs_buf": self.obs_buf}}
return observations
def calculate_metrics(self) -> None:
"""Update reward and reset buffers."""
self._update_reset_buf()
self._update_rew_buf()
def _update_reset_buf(self) -> None:
"""Assign environments for reset if successful or failed."""
# If max episode length has been reached
self.reset_buf[:] = torch.where(
self.progress_buf[:] >= self.max_episode_length - 1,
torch.ones_like(self.reset_buf),
self.reset_buf,
)
def _update_rew_buf(self) -> None:
"""Compute reward at current timestep."""
keypoint_reward = -self._get_keypoint_dist()
action_penalty = (
torch.norm(self.actions, p=2, dim=-1)
* self.cfg_task.rl.action_penalty_scale
)
self.rew_buf[:] = (
keypoint_reward * self.cfg_task.rl.keypoint_reward_scale
- action_penalty * self.cfg_task.rl.action_penalty_scale
)
# In this policy, episode length is constant across all envs
is_last_step = self.progress_buf[0] == self.max_episode_length - 1
if is_last_step:
# Check if nut is picked up and above table
lift_success = self._check_lift_success(height_multiple=3.0)
self.rew_buf[:] += lift_success * self.cfg_task.rl.success_bonus
self.extras["successes"] = torch.mean(lift_success.float())
def _get_keypoint_offsets(self, num_keypoints) -> torch.Tensor:
"""Get uniformly-spaced keypoints along a line of unit length, centered at 0."""
keypoint_offsets = torch.zeros((num_keypoints, 3), device=self.device)
keypoint_offsets[:, -1] = (
torch.linspace(0.0, 1.0, num_keypoints, device=self.device) - 0.5
)
return keypoint_offsets
def _get_keypoint_dist(self) -> torch.Tensor:
"""Get keypoint distance."""
keypoint_dist = torch.sum(
torch.norm(self.keypoints_nut - self.keypoints_gripper, p=2, dim=-1), dim=-1
)
return keypoint_dist
def _close_gripper(self, sim_steps=20) -> None:
"""Fully close gripper using controller. Called outside RL loop (i.e., after last step of episode)."""
self._move_gripper_to_dof_pos(gripper_dof_pos=0.0, sim_steps=sim_steps)
def _move_gripper_to_dof_pos(self, gripper_dof_pos, sim_steps=20) -> None:
"""Move gripper fingers to specified DOF position using controller."""
delta_hand_pose = torch.zeros(
(self.num_envs, 6), device=self.device
) # No hand motion
self._apply_actions_as_ctrl_targets(
delta_hand_pose, gripper_dof_pos, do_scale=False
)
# Step sim
for _ in range(sim_steps):
SimulationContext.step(self._env._world, render=True)
def _lift_gripper(
self, franka_gripper_width=0.0, lift_distance=0.3, sim_steps=20
) -> None:
"""Lift gripper by specified distance. Called outside RL loop (i.e., after last step of episode)."""
delta_hand_pose = torch.zeros([self.num_envs, 6], device=self.device)
delta_hand_pose[:, 2] = lift_distance
# Step sim
for _ in range(sim_steps):
self._apply_actions_as_ctrl_targets(
delta_hand_pose, franka_gripper_width, do_scale=False
)
SimulationContext.step(self._env._world, render=True)
async def _close_gripper_async(self, sim_steps=20) -> None:
"""Fully close gripper using controller. Called outside RL loop (i.e., after last step of episode)."""
await self._move_gripper_to_dof_pos_async(
gripper_dof_pos=0.0, sim_steps=sim_steps
)
async def _move_gripper_to_dof_pos_async(
self, gripper_dof_pos, sim_steps=20
) -> None:
"""Move gripper fingers to specified DOF position using controller."""
delta_hand_pose = torch.zeros(
(self.num_envs, self.cfg_task.env.numActions), device=self.device
) # No hand motion
self._apply_actions_as_ctrl_targets(
delta_hand_pose, gripper_dof_pos, do_scale=False
)
# Step sim
for _ in range(sim_steps):
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
async def _lift_gripper_async(
self, franka_gripper_width=0.0, lift_distance=0.3, sim_steps=20
) -> None:
"""Lift gripper by specified distance. Called outside RL loop (i.e., after last step of episode)."""
delta_hand_pose = torch.zeros([self.num_envs, 6], device=self.device)
delta_hand_pose[:, 2] = lift_distance
# Step sim
for _ in range(sim_steps):
self._apply_actions_as_ctrl_targets(
delta_hand_pose, franka_gripper_width, do_scale=False
)
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
def _check_lift_success(self, height_multiple) -> torch.Tensor:
"""Check if nut is above table by more than specified multiple times height of nut."""
lift_success = torch.where(
self.nut_pos[:, 2]
> self.cfg_base.env.table_height
+ self.nut_heights.squeeze(-1) * height_multiple,
torch.ones((self.num_envs,), device=self.device),
torch.zeros((self.num_envs,), device=self.device),
)
return lift_success
def _randomize_gripper_pose(self, env_ids, sim_steps) -> None:
"""Move gripper to random pose."""
# step once to update physx with the newly set joint positions from reset_franka()
SimulationContext.step(self._env._world, render=True)
# Set target pos above table
self.ctrl_target_fingertip_midpoint_pos = torch.tensor(
[0.0, 0.0, self.cfg_base.env.table_height], device=self.device
) + torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_initial, device=self.device
)
self.ctrl_target_fingertip_midpoint_pos = (
self.ctrl_target_fingertip_midpoint_pos.unsqueeze(0).repeat(
self.num_envs, 1
)
)
fingertip_midpoint_pos_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_pos_noise = fingertip_midpoint_pos_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_noise, device=self.device
)
)
self.ctrl_target_fingertip_midpoint_pos += fingertip_midpoint_pos_noise
# Set target rot
ctrl_target_fingertip_midpoint_euler = (
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_initial,
device=self.device,
)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
fingertip_midpoint_rot_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_rot_noise = fingertip_midpoint_rot_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_noise, device=self.device
)
)
ctrl_target_fingertip_midpoint_euler += fingertip_midpoint_rot_noise
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_from_euler_xyz(
ctrl_target_fingertip_midpoint_euler[:, 0],
ctrl_target_fingertip_midpoint_euler[:, 1],
ctrl_target_fingertip_midpoint_euler[:, 2],
)
# Step sim and render
for _ in range(sim_steps):
if not self._env._world.is_playing():
return
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
pos_error, axis_angle_error = fc.get_pose_error(
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
jacobian_type=self.cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_hand_pose = torch.cat((pos_error, axis_angle_error), dim=-1)
actions = torch.zeros(
(self.num_envs, self.cfg_task.env.numActions), device=self.device
)
actions[:, :6] = delta_hand_pose
self._apply_actions_as_ctrl_targets(
actions=actions,
ctrl_target_gripper_dof_pos=self.asset_info_franka_table.franka_gripper_width_max,
do_scale=False,
)
SimulationContext.step(self._env._world, render=True)
self.dof_vel[env_ids, :] = torch.zeros_like(self.dof_vel[env_ids])
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
# step once to update physx with the newly set joint velocities
SimulationContext.step(self._env._world, render=True)
async def _randomize_gripper_pose_async(self, env_ids, sim_steps) -> None:
"""Move gripper to random pose."""
# step once to update physx with the newly set joint positions from reset_franka()
await omni.kit.app.get_app().next_update_async()
# Set target pos above table
self.ctrl_target_fingertip_midpoint_pos = torch.tensor(
[0.0, 0.0, self.cfg_base.env.table_height], device=self.device
) + torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_initial, device=self.device
)
self.ctrl_target_fingertip_midpoint_pos = (
self.ctrl_target_fingertip_midpoint_pos.unsqueeze(0).repeat(
self.num_envs, 1
)
)
fingertip_midpoint_pos_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_pos_noise = fingertip_midpoint_pos_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_pos_noise, device=self.device
)
)
self.ctrl_target_fingertip_midpoint_pos += fingertip_midpoint_pos_noise
# Set target rot
ctrl_target_fingertip_midpoint_euler = (
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_initial,
device=self.device,
)
.unsqueeze(0)
.repeat(self.num_envs, 1)
)
fingertip_midpoint_rot_noise = 2 * (
torch.rand((self.num_envs, 3), dtype=torch.float32, device=self.device)
- 0.5
) # [-1, 1]
fingertip_midpoint_rot_noise = fingertip_midpoint_rot_noise @ torch.diag(
torch.tensor(
self.cfg_task.randomize.fingertip_midpoint_rot_noise, device=self.device
)
)
ctrl_target_fingertip_midpoint_euler += fingertip_midpoint_rot_noise
self.ctrl_target_fingertip_midpoint_quat = torch_utils.quat_from_euler_xyz(
ctrl_target_fingertip_midpoint_euler[:, 0],
ctrl_target_fingertip_midpoint_euler[:, 1],
ctrl_target_fingertip_midpoint_euler[:, 2],
)
# Step sim and render
for _ in range(sim_steps):
self.refresh_base_tensors()
self.refresh_env_tensors()
self._refresh_task_tensors()
pos_error, axis_angle_error = fc.get_pose_error(
fingertip_midpoint_pos=self.fingertip_midpoint_pos,
fingertip_midpoint_quat=self.fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=self.ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=self.ctrl_target_fingertip_midpoint_quat,
jacobian_type=self.cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_hand_pose = torch.cat((pos_error, axis_angle_error), dim=-1)
actions = torch.zeros(
(self.num_envs, self.cfg_task.env.numActions), device=self.device
)
actions[:, :6] = delta_hand_pose
self._apply_actions_as_ctrl_targets(
actions=actions,
ctrl_target_gripper_dof_pos=self.asset_info_franka_table.franka_gripper_width_max,
do_scale=False,
)
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
self.dof_vel[env_ids, :] = torch.zeros_like(self.dof_vel[env_ids])
indices = env_ids.to(dtype=torch.int32)
self.frankas.set_joint_velocities(self.dof_vel[env_ids], indices=indices)
# step once to update physx with the newly set joint velocities
self._env._world.physics_sim_view.flush()
await omni.kit.app.get_app().next_update_async()
| 31,568 | Python | 37.926017 | 131 | 0.589268 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_schema_class_base.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: abstract base class for base class.
Inherits ABC class. Inherited by base class. Defines template for base class.
"""
from abc import ABC, abstractmethod
class FactoryABCBase(ABC):
@abstractmethod
def __init__(self):
"""Initialize instance variables. Initialize VecTask superclass."""
pass
@abstractmethod
def _get_base_yaml_params(self):
"""Initialize instance variables from YAML files."""
pass
@abstractmethod
def import_franka_assets(self):
"""Set Franka and table asset options. Import assets."""
pass
@abstractmethod
def refresh_base_tensors(self):
"""Refresh tensors."""
# NOTE: Tensor refresh functions should be called once per step, before setters.
pass
@abstractmethod
def parse_controller_spec(self):
"""Parse controller specification into lower-level controller configuration."""
pass
@abstractmethod
def generate_ctrl_signals(self):
"""Get Jacobian. Set Franka DOF position targets or DOF torques."""
pass
@abstractmethod
def enable_gravity(self):
"""Enable gravity."""
pass
@abstractmethod
def disable_gravity(self):
"""Disable gravity."""
pass
| 2,843 | Python | 35 | 88 | 0.721069 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_schema_config_base.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: schema for base class configuration.
Used by Hydra. Defines template for base class YAML file.
"""
from dataclasses import dataclass
@dataclass
class Mode:
export_scene: bool # export scene to USD
export_states: bool # export states to NPY
@dataclass
class Sim:
dt: float # timestep size (default = 1.0 / 60.0)
num_substeps: int # number of substeps (default = 2)
num_pos_iters: int # number of position iterations for PhysX TGS solver (default = 4)
num_vel_iters: int # number of velocity iterations for PhysX TGS solver (default = 1)
gravity_mag: float # magnitude of gravitational acceleration
add_damping: bool # add damping to stabilize gripper-object interactions
@dataclass
class Env:
env_spacing: float # lateral offset between envs
franka_depth: float # depth offset of Franka base relative to env origin
table_height: float # height of table
franka_friction: float # coefficient of friction associated with Franka
table_friction: float # coefficient of friction associated with table
@dataclass
class FactorySchemaConfigBase:
mode: Mode
sim: Sim
env: Env
| 2,724 | Python | 39.073529 | 90 | 0.757342 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_env_nut_bolt.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: class for nut-bolt env.
Inherits base class and abstract environment class. Inherited by nut-bolt task classes. Not directly executed.
Configuration defined in FactoryEnvNutBolt.yaml. Asset info defined in factory_asset_info_nut_bolt.yaml.
"""
import hydra
import numpy as np
import torch
from omni.isaac.core.prims import RigidPrimView, XFormPrim
from omni.isaac.core.utils.nucleus import get_assets_root_path
from omni.isaac.core.utils.stage import add_reference_to_stage
from omniisaacgymenvs.tasks.base.rl_task import RLTask
from omni.physx.scripts import physicsUtils, utils
from omniisaacgymenvs.robots.articulations.views.factory_franka_view import (
FactoryFrankaView,
)
import omniisaacgymenvs.tasks.factory.factory_control as fc
from omniisaacgymenvs.tasks.factory.factory_base import FactoryBase
from omniisaacgymenvs.tasks.factory.factory_schema_class_env import FactoryABCEnv
from omniisaacgymenvs.tasks.factory.factory_schema_config_env import (
FactorySchemaConfigEnv,
)
class FactoryEnvNutBolt(FactoryBase, FactoryABCEnv):
def __init__(self, name, sim_config, env) -> None:
"""Initialize base superclass. Initialize instance variables."""
super().__init__(name, sim_config, env)
self._get_env_yaml_params()
def _get_env_yaml_params(self):
"""Initialize instance variables from YAML files."""
cs = hydra.core.config_store.ConfigStore.instance()
cs.store(name="factory_schema_config_env", node=FactorySchemaConfigEnv)
config_path = (
"task/FactoryEnvNutBolt.yaml" # relative to Hydra search path (cfg dir)
)
self.cfg_env = hydra.compose(config_name=config_path)
self.cfg_env = self.cfg_env["task"] # strip superfluous nesting
asset_info_path = "../tasks/factory/yaml/factory_asset_info_nut_bolt.yaml"
self.asset_info_nut_bolt = hydra.compose(config_name=asset_info_path)
self.asset_info_nut_bolt = self.asset_info_nut_bolt[""][""][""]["tasks"][
"factory"
][
"yaml"
] # strip superfluous nesting
def update_config(self, sim_config):
self._sim_config = sim_config
self._cfg = sim_config.config
self._task_cfg = sim_config.task_config
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._num_observations = self._task_cfg["env"]["numObservations"]
self._num_actions = self._task_cfg["env"]["numActions"]
self._env_spacing = self.cfg_base["env"]["env_spacing"]
self._get_env_yaml_params()
def set_up_scene(self, scene) -> None:
"""Import assets. Add to scene."""
# Increase buffer size to prevent overflow for Place and Screw tasks
physxSceneAPI = self._env._world.get_physics_context()._physx_scene_api
physxSceneAPI.CreateGpuCollisionStackSizeAttr().Set(256 * 1024 * 1024)
self.import_franka_assets(add_to_stage=True)
self.create_nut_bolt_material()
RLTask.set_up_scene(self, scene, replicate_physics=False)
self._import_env_assets(add_to_stage=True)
self.frankas = FactoryFrankaView(
prim_paths_expr="/World/envs/.*/franka", name="frankas_view"
)
self.nuts = RigidPrimView(
prim_paths_expr="/World/envs/.*/nut/factory_nut.*",
name="nuts_view",
track_contact_forces=True,
)
self.bolts = RigidPrimView(
prim_paths_expr="/World/envs/.*/bolt/factory_bolt.*",
name="bolts_view",
track_contact_forces=True,
)
scene.add(self.nuts)
scene.add(self.bolts)
scene.add(self.frankas)
scene.add(self.frankas._hands)
scene.add(self.frankas._lfingers)
scene.add(self.frankas._rfingers)
scene.add(self.frankas._fingertip_centered)
return
def initialize_views(self, scene) -> None:
"""Initialize views for extension workflow."""
super().initialize_views(scene)
self.import_franka_assets(add_to_stage=False)
self._import_env_assets(add_to_stage=False)
if scene.object_exists("frankas_view"):
scene.remove_object("frankas_view", registry_only=True)
if scene.object_exists("nuts_view"):
scene.remove_object("nuts_view", registry_only=True)
if scene.object_exists("bolts_view"):
scene.remove_object("bolts_view", registry_only=True)
if scene.object_exists("hands_view"):
scene.remove_object("hands_view", registry_only=True)
if scene.object_exists("lfingers_view"):
scene.remove_object("lfingers_view", registry_only=True)
if scene.object_exists("rfingers_view"):
scene.remove_object("rfingers_view", registry_only=True)
if scene.object_exists("fingertips_view"):
scene.remove_object("fingertips_view", registry_only=True)
self.frankas = FactoryFrankaView(
prim_paths_expr="/World/envs/.*/franka", name="frankas_view"
)
self.nuts = RigidPrimView(
prim_paths_expr="/World/envs/.*/nut/factory_nut.*", name="nuts_view"
)
self.bolts = RigidPrimView(
prim_paths_expr="/World/envs/.*/bolt/factory_bolt.*", name="bolts_view"
)
scene.add(self.nuts)
scene.add(self.bolts)
scene.add(self.frankas)
scene.add(self.frankas._hands)
scene.add(self.frankas._lfingers)
scene.add(self.frankas._rfingers)
scene.add(self.frankas._fingertip_centered)
def create_nut_bolt_material(self):
"""Define nut and bolt material."""
self.nutboltPhysicsMaterialPath = "/World/Physics_Materials/NutBoltMaterial"
utils.addRigidBodyMaterial(
self._stage,
self.nutboltPhysicsMaterialPath,
density=self.cfg_env.env.nut_bolt_density,
staticFriction=self.cfg_env.env.nut_bolt_friction,
dynamicFriction=self.cfg_env.env.nut_bolt_friction,
restitution=0.0,
)
def _import_env_assets(self, add_to_stage=True):
"""Set nut and bolt asset options. Import assets."""
self.nut_heights = []
self.nut_widths_max = []
self.bolt_widths = []
self.bolt_head_heights = []
self.bolt_shank_lengths = []
self.thread_pitches = []
assets_root_path = get_assets_root_path()
for i in range(0, self._num_envs):
j = np.random.randint(0, len(self.cfg_env.env.desired_subassemblies))
subassembly = self.cfg_env.env.desired_subassemblies[j]
components = list(self.asset_info_nut_bolt[subassembly])
nut_translation = torch.tensor(
[
0.0,
self.cfg_env.env.nut_lateral_offset,
self.cfg_base.env.table_height,
],
device=self._device,
)
nut_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self._device)
nut_height = self.asset_info_nut_bolt[subassembly][components[0]]["height"]
nut_width_max = self.asset_info_nut_bolt[subassembly][components[0]][
"width_max"
]
self.nut_heights.append(nut_height)
self.nut_widths_max.append(nut_width_max)
nut_file = (
assets_root_path
+ self.asset_info_nut_bolt[subassembly][components[0]]["usd_path"]
)
if add_to_stage:
add_reference_to_stage(nut_file, f"/World/envs/env_{i}" + "/nut")
XFormPrim(
prim_path=f"/World/envs/env_{i}" + "/nut",
translation=nut_translation,
orientation=nut_orientation,
)
self._stage.GetPrimAtPath(
f"/World/envs/env_{i}" + f"/nut/factory_{components[0]}/collisions"
).SetInstanceable(
False
) # This is required to be able to edit physics material
physicsUtils.add_physics_material_to_prim(
self._stage,
self._stage.GetPrimAtPath(
f"/World/envs/env_{i}"
+ f"/nut/factory_{components[0]}/collisions/mesh_0"
),
self.nutboltPhysicsMaterialPath,
)
# applies articulation settings from the task configuration yaml file
self._sim_config.apply_articulation_settings(
"nut",
self._stage.GetPrimAtPath(f"/World/envs/env_{i}" + "/nut"),
self._sim_config.parse_actor_config("nut"),
)
bolt_translation = torch.tensor(
[0.0, 0.0, self.cfg_base.env.table_height], device=self._device
)
bolt_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self._device)
bolt_width = self.asset_info_nut_bolt[subassembly][components[1]]["width"]
bolt_head_height = self.asset_info_nut_bolt[subassembly][components[1]][
"head_height"
]
bolt_shank_length = self.asset_info_nut_bolt[subassembly][components[1]][
"shank_length"
]
self.bolt_widths.append(bolt_width)
self.bolt_head_heights.append(bolt_head_height)
self.bolt_shank_lengths.append(bolt_shank_length)
if add_to_stage:
bolt_file = (
assets_root_path
+ self.asset_info_nut_bolt[subassembly][components[1]]["usd_path"]
)
add_reference_to_stage(bolt_file, f"/World/envs/env_{i}" + "/bolt")
XFormPrim(
prim_path=f"/World/envs/env_{i}" + "/bolt",
translation=bolt_translation,
orientation=bolt_orientation,
)
self._stage.GetPrimAtPath(
f"/World/envs/env_{i}" + f"/bolt/factory_{components[1]}/collisions"
).SetInstanceable(
False
) # This is required to be able to edit physics material
physicsUtils.add_physics_material_to_prim(
self._stage,
self._stage.GetPrimAtPath(
f"/World/envs/env_{i}"
+ f"/bolt/factory_{components[1]}/collisions/mesh_0"
),
self.nutboltPhysicsMaterialPath,
)
# applies articulation settings from the task configuration yaml file
self._sim_config.apply_articulation_settings(
"bolt",
self._stage.GetPrimAtPath(f"/World/envs/env_{i}" + "/bolt"),
self._sim_config.parse_actor_config("bolt"),
)
thread_pitch = self.asset_info_nut_bolt[subassembly]["thread_pitch"]
self.thread_pitches.append(thread_pitch)
# For computing body COM pos
self.nut_heights = torch.tensor(
self.nut_heights, device=self._device
).unsqueeze(-1)
self.bolt_head_heights = torch.tensor(
self.bolt_head_heights, device=self._device
).unsqueeze(-1)
# For setting initial state
self.nut_widths_max = torch.tensor(
self.nut_widths_max, device=self._device
).unsqueeze(-1)
self.bolt_shank_lengths = torch.tensor(
self.bolt_shank_lengths, device=self._device
).unsqueeze(-1)
# For defining success or failure
self.bolt_widths = torch.tensor(
self.bolt_widths, device=self._device
).unsqueeze(-1)
self.thread_pitches = torch.tensor(
self.thread_pitches, device=self._device
).unsqueeze(-1)
def refresh_env_tensors(self):
"""Refresh tensors."""
# Nut tensors
self.nut_pos, self.nut_quat = self.nuts.get_world_poses(clone=False)
self.nut_pos -= self.env_pos
self.nut_com_pos = fc.translate_along_local_z(
pos=self.nut_pos,
quat=self.nut_quat,
offset=self.bolt_head_heights + self.nut_heights * 0.5,
device=self.device,
)
self.nut_com_quat = self.nut_quat # always equal
nut_velocities = self.nuts.get_velocities(clone=False)
self.nut_linvel = nut_velocities[:, 0:3]
self.nut_angvel = nut_velocities[:, 3:6]
self.nut_com_linvel = self.nut_linvel + torch.cross(
self.nut_angvel, (self.nut_com_pos - self.nut_pos), dim=1
)
self.nut_com_angvel = self.nut_angvel # always equal
self.nut_force = self.nuts.get_net_contact_forces(clone=False)
# Bolt tensors
self.bolt_pos, self.bolt_quat = self.bolts.get_world_poses(clone=False)
self.bolt_pos -= self.env_pos
self.bolt_force = self.bolts.get_net_contact_forces(clone=False)
| 14,709 | Python | 39.30137 | 110 | 0.603372 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/factory_control.py | # Copyright (c) 2018-2023, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Factory: control module.
Imported by base, environment, and task classes. Not directly executed.
"""
import math
import omni.isaac.core.utils.torch as torch_utils
import torch
def compute_dof_pos_target(
cfg_ctrl,
arm_dof_pos,
fingertip_midpoint_pos,
fingertip_midpoint_quat,
jacobian,
ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat,
ctrl_target_gripper_dof_pos,
device,
):
"""Compute Franka DOF position target to move fingertips towards target pose."""
ctrl_target_dof_pos = torch.zeros((cfg_ctrl["num_envs"], 9), device=device)
pos_error, axis_angle_error = get_pose_error(
fingertip_midpoint_pos=fingertip_midpoint_pos,
fingertip_midpoint_quat=fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=ctrl_target_fingertip_midpoint_quat,
jacobian_type=cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_fingertip_pose = torch.cat((pos_error, axis_angle_error), dim=1)
delta_arm_dof_pos = _get_delta_dof_pos(
delta_pose=delta_fingertip_pose,
ik_method=cfg_ctrl["ik_method"],
jacobian=jacobian,
device=device,
)
ctrl_target_dof_pos[:, 0:7] = arm_dof_pos + delta_arm_dof_pos
ctrl_target_dof_pos[:, 7:9] = ctrl_target_gripper_dof_pos # gripper finger joints
return ctrl_target_dof_pos
def compute_dof_torque(
cfg_ctrl,
dof_pos,
dof_vel,
fingertip_midpoint_pos,
fingertip_midpoint_quat,
fingertip_midpoint_linvel,
fingertip_midpoint_angvel,
left_finger_force,
right_finger_force,
jacobian,
arm_mass_matrix,
ctrl_target_gripper_dof_pos,
ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat,
ctrl_target_fingertip_contact_wrench,
device,
):
"""Compute Franka DOF torque to move fingertips towards target pose."""
# References:
# 1) https://ethz.ch/content/dam/ethz/special-interest/mavt/robotics-n-intelligent-systems/rsl-dam/documents/RobotDynamics2018/RD_HS2018script.pdf
# 2) Modern Robotics
dof_torque = torch.zeros((cfg_ctrl["num_envs"], 9), device=device)
if cfg_ctrl["gain_space"] == "joint":
pos_error, axis_angle_error = get_pose_error(
fingertip_midpoint_pos=fingertip_midpoint_pos,
fingertip_midpoint_quat=fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=ctrl_target_fingertip_midpoint_quat,
jacobian_type=cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_fingertip_pose = torch.cat((pos_error, axis_angle_error), dim=1)
# Set tau = k_p * joint_pos_error - k_d * joint_vel_error (ETH eq. 3.72)
delta_arm_dof_pos = _get_delta_dof_pos(
delta_pose=delta_fingertip_pose,
ik_method=cfg_ctrl["ik_method"],
jacobian=jacobian,
device=device,
)
dof_torque[:, 0:7] = cfg_ctrl[
"joint_prop_gains"
] * delta_arm_dof_pos + cfg_ctrl["joint_deriv_gains"] * (0.0 - dof_vel[:, 0:7])
if cfg_ctrl["do_inertial_comp"]:
# Set tau = M * tau, where M is the joint-space mass matrix
arm_mass_matrix_joint = arm_mass_matrix
dof_torque[:, 0:7] = (
arm_mass_matrix_joint @ dof_torque[:, 0:7].unsqueeze(-1)
).squeeze(-1)
elif cfg_ctrl["gain_space"] == "task":
task_wrench = torch.zeros((cfg_ctrl["num_envs"], 6), device=device)
if cfg_ctrl["do_motion_ctrl"]:
pos_error, axis_angle_error = get_pose_error(
fingertip_midpoint_pos=fingertip_midpoint_pos,
fingertip_midpoint_quat=fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos=ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat=ctrl_target_fingertip_midpoint_quat,
jacobian_type=cfg_ctrl["jacobian_type"],
rot_error_type="axis_angle",
)
delta_fingertip_pose = torch.cat((pos_error, axis_angle_error), dim=1)
# Set tau = k_p * task_pos_error - k_d * task_vel_error (building towards eq. 3.96-3.98)
task_wrench_motion = _apply_task_space_gains(
delta_fingertip_pose=delta_fingertip_pose,
fingertip_midpoint_linvel=fingertip_midpoint_linvel,
fingertip_midpoint_angvel=fingertip_midpoint_angvel,
task_prop_gains=cfg_ctrl["task_prop_gains"],
task_deriv_gains=cfg_ctrl["task_deriv_gains"],
)
if cfg_ctrl["do_inertial_comp"]:
# Set tau = Lambda * tau, where Lambda is the task-space mass matrix
jacobian_T = torch.transpose(jacobian, dim0=1, dim1=2)
arm_mass_matrix_task = torch.inverse(
jacobian @ torch.inverse(arm_mass_matrix) @ jacobian_T
) # ETH eq. 3.86; geometric Jacobian is assumed
task_wrench_motion = (
arm_mass_matrix_task @ task_wrench_motion.unsqueeze(-1)
).squeeze(-1)
task_wrench = (
task_wrench + cfg_ctrl["motion_ctrl_axes"] * task_wrench_motion
)
if cfg_ctrl["do_force_ctrl"]:
# Set tau = tau + F_t, where F_t is the target contact wrench
task_wrench_force = torch.zeros((cfg_ctrl["num_envs"], 6), device=device)
task_wrench_force = (
task_wrench_force + ctrl_target_fingertip_contact_wrench
) # open-loop force control (building towards ETH eq. 3.96-3.98)
if cfg_ctrl["force_ctrl_method"] == "closed":
force_error, torque_error = _get_wrench_error(
left_finger_force=left_finger_force,
right_finger_force=right_finger_force,
ctrl_target_fingertip_contact_wrench=ctrl_target_fingertip_contact_wrench,
num_envs=cfg_ctrl["num_envs"],
device=device,
)
# Set tau = tau + k_p * contact_wrench_error
task_wrench_force = task_wrench_force + cfg_ctrl[
"wrench_prop_gains"
] * torch.cat(
(force_error, torque_error), dim=1
) # part of Modern Robotics eq. 11.61
task_wrench = (
task_wrench
+ torch.tensor(cfg_ctrl["force_ctrl_axes"], device=device).unsqueeze(0)
* task_wrench_force
)
# Set tau = J^T * tau, i.e., map tau into joint space as desired
jacobian_T = torch.transpose(jacobian, dim0=1, dim1=2)
dof_torque[:, 0:7] = (jacobian_T @ task_wrench.unsqueeze(-1)).squeeze(-1)
dof_torque[:, 7:9] = cfg_ctrl["gripper_prop_gains"] * (
ctrl_target_gripper_dof_pos - dof_pos[:, 7:9]
) + cfg_ctrl["gripper_deriv_gains"] * (
0.0 - dof_vel[:, 7:9]
) # gripper finger joints
dof_torque = torch.clamp(dof_torque, min=-100.0, max=100.0)
return dof_torque
def get_pose_error(
fingertip_midpoint_pos,
fingertip_midpoint_quat,
ctrl_target_fingertip_midpoint_pos,
ctrl_target_fingertip_midpoint_quat,
jacobian_type,
rot_error_type,
):
"""Compute task-space error between target Franka fingertip pose and current pose."""
# Reference: https://ethz.ch/content/dam/ethz/special-interest/mavt/robotics-n-intelligent-systems/rsl-dam/documents/RobotDynamics2018/RD_HS2018script.pdf
# Compute pos error
pos_error = ctrl_target_fingertip_midpoint_pos - fingertip_midpoint_pos
# Compute rot error
if (
jacobian_type == "geometric"
): # See example 2.9.8; note use of J_g and transformation between rotation vectors
# Compute quat error (i.e., difference quat)
# Reference: https://personal.utdallas.edu/~sxb027100/dock/quat.html
fingertip_midpoint_quat_norm = torch_utils.quat_mul(
fingertip_midpoint_quat, torch_utils.quat_conjugate(fingertip_midpoint_quat)
)[
:, 0
] # scalar component
fingertip_midpoint_quat_inv = torch_utils.quat_conjugate(
fingertip_midpoint_quat
) / fingertip_midpoint_quat_norm.unsqueeze(-1)
quat_error = torch_utils.quat_mul(
ctrl_target_fingertip_midpoint_quat, fingertip_midpoint_quat_inv
)
# Convert to axis-angle error
axis_angle_error = axis_angle_from_quat(quat_error)
elif (
jacobian_type == "analytic"
): # See example 2.9.7; note use of J_a and difference of rotation vectors
# Compute axis-angle error
axis_angle_error = axis_angle_from_quat(
ctrl_target_fingertip_midpoint_quat
) - axis_angle_from_quat(fingertip_midpoint_quat)
if rot_error_type == "quat":
return pos_error, quat_error
elif rot_error_type == "axis_angle":
return pos_error, axis_angle_error
def _get_wrench_error(
left_finger_force,
right_finger_force,
ctrl_target_fingertip_contact_wrench,
num_envs,
device,
):
"""Compute task-space error between target Franka fingertip contact wrench and current wrench."""
fingertip_contact_wrench = torch.zeros((num_envs, 6), device=device)
fingertip_contact_wrench[:, 0:3] = (
left_finger_force + right_finger_force
) # net contact force on fingers
# Cols 3 to 6 are all zeros, as we do not have enough information
force_error = ctrl_target_fingertip_contact_wrench[:, 0:3] - (
-fingertip_contact_wrench[:, 0:3]
)
torque_error = ctrl_target_fingertip_contact_wrench[:, 3:6] - (
-fingertip_contact_wrench[:, 3:6]
)
return force_error, torque_error
def _get_delta_dof_pos(delta_pose, ik_method, jacobian, device):
"""Get delta Franka DOF position from delta pose using specified IK method."""
# References:
# 1) https://www.cs.cmu.edu/~15464-s13/lectures/lecture6/iksurvey.pdf
# 2) https://ethz.ch/content/dam/ethz/special-interest/mavt/robotics-n-intelligent-systems/rsl-dam/documents/RobotDynamics2018/RD_HS2018script.pdf (p. 47)
if ik_method == "pinv": # Jacobian pseudoinverse
k_val = 1.0
jacobian_pinv = torch.linalg.pinv(jacobian)
delta_dof_pos = k_val * jacobian_pinv @ delta_pose.unsqueeze(-1)
delta_dof_pos = delta_dof_pos.squeeze(-1)
elif ik_method == "trans": # Jacobian transpose
k_val = 1.0
jacobian_T = torch.transpose(jacobian, dim0=1, dim1=2)
delta_dof_pos = k_val * jacobian_T @ delta_pose.unsqueeze(-1)
delta_dof_pos = delta_dof_pos.squeeze(-1)
elif ik_method == "dls": # damped least squares (Levenberg-Marquardt)
lambda_val = 0.1
jacobian_T = torch.transpose(jacobian, dim0=1, dim1=2)
lambda_matrix = (lambda_val**2) * torch.eye(
n=jacobian.shape[1], device=device
)
delta_dof_pos = (
jacobian_T
@ torch.inverse(jacobian @ jacobian_T + lambda_matrix)
@ delta_pose.unsqueeze(-1)
)
delta_dof_pos = delta_dof_pos.squeeze(-1)
elif ik_method == "svd": # adaptive SVD
k_val = 1.0
U, S, Vh = torch.linalg.svd(jacobian)
S_inv = 1.0 / S
min_singular_value = 1.0e-5
S_inv = torch.where(S > min_singular_value, S_inv, torch.zeros_like(S_inv))
jacobian_pinv = (
torch.transpose(Vh, dim0=1, dim1=2)[:, :, :6]
@ torch.diag_embed(S_inv)
@ torch.transpose(U, dim0=1, dim1=2)
)
delta_dof_pos = k_val * jacobian_pinv @ delta_pose.unsqueeze(-1)
delta_dof_pos = delta_dof_pos.squeeze(-1)
return delta_dof_pos
def _apply_task_space_gains(
delta_fingertip_pose,
fingertip_midpoint_linvel,
fingertip_midpoint_angvel,
task_prop_gains,
task_deriv_gains,
):
"""Interpret PD gains as task-space gains. Apply to task-space error."""
task_wrench = torch.zeros_like(delta_fingertip_pose)
# Apply gains to lin error components
lin_error = delta_fingertip_pose[:, 0:3]
task_wrench[:, 0:3] = task_prop_gains[:, 0:3] * lin_error + task_deriv_gains[
:, 0:3
] * (0.0 - fingertip_midpoint_linvel)
# Apply gains to rot error components
rot_error = delta_fingertip_pose[:, 3:6]
task_wrench[:, 3:6] = task_prop_gains[:, 3:6] * rot_error + task_deriv_gains[
:, 3:6
] * (0.0 - fingertip_midpoint_angvel)
return task_wrench
def get_analytic_jacobian(fingertip_quat, fingertip_jacobian, num_envs, device):
"""Convert geometric Jacobian to analytic Jacobian."""
# Reference: https://ethz.ch/content/dam/ethz/special-interest/mavt/robotics-n-intelligent-systems/rsl-dam/documents/RobotDynamics2018/RD_HS2018script.pdf
# NOTE: Gym returns world-space geometric Jacobians by default
batch = num_envs
# Overview:
# x = [x_p; x_r]
# From eq. 2.189 and 2.192, x_dot = J_a @ q_dot = (E_inv @ J_g) @ q_dot
# From eq. 2.191, E = block(E_p, E_r); thus, E_inv = block(E_p_inv, E_r_inv)
# Eq. 2.12 gives an expression for E_p_inv
# Eq. 2.107 gives an expression for E_r_inv
# Compute E_inv_top (i.e., [E_p_inv, 0])
I = torch.eye(3, device=device)
E_p_inv = I.repeat((batch, 1)).reshape(batch, 3, 3)
E_inv_top = torch.cat((E_p_inv, torch.zeros((batch, 3, 3), device=device)), dim=2)
# Compute E_inv_bottom (i.e., [0, E_r_inv])
fingertip_axis_angle = axis_angle_from_quat(fingertip_quat)
fingertip_axis_angle_cross = get_skew_symm_matrix(
fingertip_axis_angle, device=device
)
fingertip_angle = torch.linalg.vector_norm(fingertip_axis_angle, dim=1)
factor_1 = 1 / (fingertip_angle**2)
factor_2 = 1 - fingertip_angle * 0.5 * torch.sin(fingertip_angle) / (
1 - torch.cos(fingertip_angle)
)
factor_3 = factor_1 * factor_2
E_r_inv = (
I
- 1 * 0.5 * fingertip_axis_angle_cross
+ (fingertip_axis_angle_cross @ fingertip_axis_angle_cross)
* factor_3.unsqueeze(-1).repeat((1, 3 * 3)).reshape((batch, 3, 3))
)
E_inv_bottom = torch.cat(
(torch.zeros((batch, 3, 3), device=device), E_r_inv), dim=2
)
E_inv = torch.cat(
(E_inv_top.reshape((batch, 3 * 6)), E_inv_bottom.reshape((batch, 3 * 6))), dim=1
).reshape((batch, 6, 6))
J_a = E_inv @ fingertip_jacobian
return J_a
def get_skew_symm_matrix(vec, device):
"""Convert vector to skew-symmetric matrix."""
# Reference: https://en.wikipedia.org/wiki/Cross_product#Conversion_to_matrix_multiplication
batch = vec.shape[0]
I = torch.eye(3, device=device)
skew_symm = torch.transpose(
torch.cross(
vec.repeat((1, 3)).reshape((batch * 3, 3)), I.repeat((batch, 1))
).reshape(batch, 3, 3),
dim0=1,
dim1=2,
)
return skew_symm
def translate_along_local_z(pos, quat, offset, device):
"""Translate global body position along local Z-axis and express in global coordinates."""
num_vecs = pos.shape[0]
offset_vec = offset * torch.tensor([0.0, 0.0, 1.0], device=device).repeat(
(num_vecs, 1)
)
_, translated_pos = torch_utils.tf_combine(
q1=quat,
t1=pos,
q2=torch.tensor([1.0, 0.0, 0.0, 0.0], device=device).repeat((num_vecs, 1)),
t2=offset_vec,
)
return translated_pos
def axis_angle_from_euler(euler):
"""Convert tensor of Euler angles to tensor of axis-angles."""
quat = torch_utils.quat_from_euler_xyz(
roll=euler[:, 0], pitch=euler[:, 1], yaw=euler[:, 2]
)
quat = quat * torch.sign(quat[:, 0]).unsqueeze(-1) # smaller rotation
axis_angle = axis_angle_from_quat(quat)
return axis_angle
def axis_angle_from_quat(quat, eps=1.0e-6):
"""Convert tensor of quaternions to tensor of axis-angles."""
# Reference: https://github.com/facebookresearch/pytorch3d/blob/bee31c48d3d36a8ea268f9835663c52ff4a476ec/pytorch3d/transforms/rotation_conversions.py#L516-L544
mag = torch.linalg.norm(quat[:, 1:4], dim=1)
half_angle = torch.atan2(mag, quat[:, 0])
angle = 2.0 * half_angle
sin_half_angle_over_angle = torch.where(
torch.abs(angle) > eps, torch.sin(half_angle) / angle, 1 / 2 - angle**2.0 / 48
)
axis_angle = quat[:, 1:4] / sin_half_angle_over_angle.unsqueeze(-1)
return axis_angle
def axis_angle_from_quat_naive(quat):
"""Convert tensor of quaternions to tensor of axis-angles."""
# Reference: https://en.wikipedia.org/wiki/quats_and_spatial_rotation#Recovering_the_axis-angle_representation
# NOTE: Susceptible to undesirable behavior due to divide-by-zero
mag = torch.linalg.vector_norm(quat[:, 1:4], dim=1) # zero when quat = [1, 0, 0, 0]
axis = quat[:, 1:4] / mag.unsqueeze(-1)
angle = 2.0 * torch.atan2(mag, quat[:, 0])
axis_angle = axis * angle.unsqueeze(-1)
return axis_angle
def get_rand_quat(num_quats, device):
"""Generate tensor of random quaternions."""
# Reference: http://planning.cs.uiuc.edu/node198.html
u = torch.rand((num_quats, 3), device=device)
quat = torch.zeros((num_quats, 4), device=device)
quat[:, 0] = torch.sqrt(u[:, 0]) * torch.cos(2 * math.pi * u[:, 2])
quat[:, 1] = torch.sqrt(1 - u[:, 0]) * torch.sin(2 * math.pi * u[:, 1])
quat[:, 2] = torch.sqrt(1 - u[:, 0]) * torch.cos(2 * math.pi * u[:, 1])
quat[:, 3] = torch.sqrt(u[:, 0]) * torch.sin(2 * math.pi * u[:, 2])
return quat
def get_nonrand_quat(num_quats, rot_perturbation, device):
"""Generate tensor of non-random quaternions by composing random Euler rotations."""
quat = torch_utils.quat_from_euler_xyz(
torch.rand((num_quats, 1), device=device).squeeze() * rot_perturbation * 2.0
- rot_perturbation,
torch.rand((num_quats, 1), device=device).squeeze() * rot_perturbation * 2.0
- rot_perturbation,
torch.rand((num_quats, 1), device=device).squeeze() * rot_perturbation * 2.0
- rot_perturbation,
)
return quat
| 19,859 | Python | 37.864971 | 163 | 0.627574 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/yaml/factory_asset_info_nut_bolt.yaml | nut_bolt_m4:
nut:
usd_path: '/Isaac/Props/Factory/factory_nut_m4_tight/factory_nut_m4_tight.usd'
width_min: 0.007 # distance from flat surface to flat surface
width_max: 0.0080829 # distance from edge to edge
height: 0.0032 # height of nut
flat_length: 0.00404145 # length of flat surface
bolt:
usd_path: '/Isaac/Props/Factory/factory_bolt_m4_tight/factory_bolt_m4_tight.usd'
width: 0.004 # major diameter of bolt
head_height: 0.004 # height of bolt head
shank_length: 0.016 # length of bolt shank
thread_pitch: 0.0007 # distance between threads
nut_bolt_m8:
nut:
usd_path: '/Isaac/Props/Factory/factory_nut_m8_tight/factory_nut_m8_tight.usd'
width_min: 0.013
width_max: 0.01501111
height: 0.0065
flat_length: 0.00750555
bolt:
usd_path: '/Isaac/Props/Factory/factory_bolt_m8_tight/factory_bolt_m8_tight.usd'
width: 0.008
head_height: 0.008
shank_length: 0.018
thread_pitch: 0.00125
nut_bolt_m12:
nut:
usd_path: '/Isaac/Props/Factory/factory_nut_m12_tight/factory_nut_m12_tight.usd'
width_min: 0.019
width_max: 0.02193931
height: 0.010
flat_length: 0.01096966
bolt:
usd_path: '/Isaac/Props/Factory/factory_bolt_m12_tight/factory_bolt_m12_tight.usd'
width: 0.012
head_height: 0.012
shank_length: 0.020
thread_pitch: 0.00175
nut_bolt_m16:
nut:
usd_path: '/Isaac/Props/Factory/factory_nut_m16_tight/factory_nut_m16_tight.usd'
width_min: 0.024
width_max: 0.02771281
height: 0.013
flat_length: 0.01385641
bolt:
usd_path: '/Isaac/Props/Factory/factory_bolt_m16_tight/factory_bolt_m16_tight.usd'
width: 0.016
head_height: 0.016
shank_length: 0.025
thread_pitch: 0.002
nut_bolt_m20:
nut:
usd_path: '/Isaac/Props/Factory/factory_nut_m20_tight/factory_nut_m20_tight.usd'
width_min: 0.030
width_max: 0.03464102
height: 0.016
flat_length: 0.01732051
bolt:
usd_path: '/Isaac/Props/Factory/factory_bolt_m20_tight/factory_bolt_m20_tight.usd'
width: 0.020
head_height: 0.020
shank_length: 0.045
thread_pitch: 0.0025
| 2,331 | YAML | 32.314285 | 90 | 0.617332 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/factory/yaml/factory_asset_info_franka_table.yaml | franka_hand_length: 0.0584 # distance from origin of hand to origin of finger
franka_finger_length: 0.053671 # distance from origin of finger to bottom of fingerpad
franka_fingerpad_length: 0.017608 # distance from top of inner surface of fingerpad to bottom of inner surface of fingerpad
franka_gripper_width_max: 0.080 # maximum opening width of gripper
table_depth: 0.6 # depth of table
table_width: 1.0 # width of table
| 431 | YAML | 52.999993 | 124 | 0.772622 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/utils/anymal_terrain_generator.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
import numpy as np
import torch
from omniisaacgymenvs.utils.terrain_utils.terrain_utils import *
# terrain generator
class Terrain:
def __init__(self, cfg, num_robots) -> None:
self.horizontal_scale = 0.1
self.vertical_scale = 0.005
self.border_size = 20
self.num_per_env = 2
self.env_length = cfg["mapLength"]
self.env_width = cfg["mapWidth"]
self.proportions = [np.sum(cfg["terrainProportions"][: i + 1]) for i in range(len(cfg["terrainProportions"]))]
self.env_rows = cfg["numLevels"]
self.env_cols = cfg["numTerrains"]
self.num_maps = self.env_rows * self.env_cols
self.num_per_env = int(num_robots / self.num_maps)
self.env_origins = np.zeros((self.env_rows, self.env_cols, 3))
self.width_per_env_pixels = int(self.env_width / self.horizontal_scale)
self.length_per_env_pixels = int(self.env_length / self.horizontal_scale)
self.border = int(self.border_size / self.horizontal_scale)
self.tot_cols = int(self.env_cols * self.width_per_env_pixels) + 2 * self.border
self.tot_rows = int(self.env_rows * self.length_per_env_pixels) + 2 * self.border
self.height_field_raw = np.zeros((self.tot_rows, self.tot_cols), dtype=np.int16)
if cfg["curriculum"]:
self.curiculum(num_robots, num_terrains=self.env_cols, num_levels=self.env_rows)
else:
self.randomized_terrain()
self.heightsamples = self.height_field_raw
self.vertices, self.triangles = convert_heightfield_to_trimesh(
self.height_field_raw, self.horizontal_scale, self.vertical_scale, cfg["slopeTreshold"]
)
def randomized_terrain(self):
for k in range(self.num_maps):
# Env coordinates in the world
(i, j) = np.unravel_index(k, (self.env_rows, self.env_cols))
# Heightfield coordinate system from now on
start_x = self.border + i * self.length_per_env_pixels
end_x = self.border + (i + 1) * self.length_per_env_pixels
start_y = self.border + j * self.width_per_env_pixels
end_y = self.border + (j + 1) * self.width_per_env_pixels
terrain = SubTerrain(
"terrain",
width=self.width_per_env_pixels,
length=self.width_per_env_pixels,
vertical_scale=self.vertical_scale,
horizontal_scale=self.horizontal_scale,
)
choice = np.random.uniform(0, 1)
if choice < 0.1:
if np.random.choice([0, 1]):
pyramid_sloped_terrain(terrain, np.random.choice([-0.3, -0.2, 0, 0.2, 0.3]))
random_uniform_terrain(terrain, min_height=-0.1, max_height=0.1, step=0.05, downsampled_scale=0.2)
else:
pyramid_sloped_terrain(terrain, np.random.choice([-0.3, -0.2, 0, 0.2, 0.3]))
elif choice < 0.6:
# step_height = np.random.choice([-0.18, -0.15, -0.1, -0.05, 0.05, 0.1, 0.15, 0.18])
step_height = np.random.choice([-0.15, 0.15])
pyramid_stairs_terrain(terrain, step_width=0.31, step_height=step_height, platform_size=3.0)
elif choice < 1.0:
discrete_obstacles_terrain(terrain, 0.15, 1.0, 2.0, 40, platform_size=3.0)
self.height_field_raw[start_x:end_x, start_y:end_y] = terrain.height_field_raw
env_origin_x = (i + 0.5) * self.env_length
env_origin_y = (j + 0.5) * self.env_width
x1 = int((self.env_length / 2.0 - 1) / self.horizontal_scale)
x2 = int((self.env_length / 2.0 + 1) / self.horizontal_scale)
y1 = int((self.env_width / 2.0 - 1) / self.horizontal_scale)
y2 = int((self.env_width / 2.0 + 1) / self.horizontal_scale)
env_origin_z = np.max(terrain.height_field_raw[x1:x2, y1:y2]) * self.vertical_scale
self.env_origins[i, j] = [env_origin_x, env_origin_y, env_origin_z]
def curiculum(self, num_robots, num_terrains, num_levels):
num_robots_per_map = int(num_robots / num_terrains)
left_over = num_robots % num_terrains
idx = 0
for j in range(num_terrains):
for i in range(num_levels):
terrain = SubTerrain(
"terrain",
width=self.width_per_env_pixels,
length=self.width_per_env_pixels,
vertical_scale=self.vertical_scale,
horizontal_scale=self.horizontal_scale,
)
difficulty = i / num_levels
choice = j / num_terrains
slope = difficulty * 0.4
step_height = 0.05 + 0.175 * difficulty
discrete_obstacles_height = 0.025 + difficulty * 0.15
stepping_stones_size = 2 - 1.8 * difficulty
if choice < self.proportions[0]:
if choice < 0.05:
slope *= -1
pyramid_sloped_terrain(terrain, slope=slope, platform_size=3.0)
elif choice < self.proportions[1]:
if choice < 0.15:
slope *= -1
pyramid_sloped_terrain(terrain, slope=slope, platform_size=3.0)
random_uniform_terrain(terrain, min_height=-0.1, max_height=0.1, step=0.025, downsampled_scale=0.2)
elif choice < self.proportions[3]:
if choice < self.proportions[2]:
step_height *= -1
pyramid_stairs_terrain(terrain, step_width=0.31, step_height=step_height, platform_size=3.0)
elif choice < self.proportions[4]:
discrete_obstacles_terrain(terrain, discrete_obstacles_height, 1.0, 2.0, 40, platform_size=3.0)
else:
stepping_stones_terrain(
terrain, stone_size=stepping_stones_size, stone_distance=0.1, max_height=0.0, platform_size=3.0
)
# Heightfield coordinate system
start_x = self.border + i * self.length_per_env_pixels
end_x = self.border + (i + 1) * self.length_per_env_pixels
start_y = self.border + j * self.width_per_env_pixels
end_y = self.border + (j + 1) * self.width_per_env_pixels
self.height_field_raw[start_x:end_x, start_y:end_y] = terrain.height_field_raw
robots_in_map = num_robots_per_map
if j < left_over:
robots_in_map += 1
env_origin_x = (i + 0.5) * self.env_length
env_origin_y = (j + 0.5) * self.env_width
x1 = int((self.env_length / 2.0 - 1) / self.horizontal_scale)
x2 = int((self.env_length / 2.0 + 1) / self.horizontal_scale)
y1 = int((self.env_width / 2.0 - 1) / self.horizontal_scale)
y2 = int((self.env_width / 2.0 + 1) / self.horizontal_scale)
env_origin_z = np.max(terrain.height_field_raw[x1:x2, y1:y2]) * self.vertical_scale
self.env_origins[i, j] = [env_origin_x, env_origin_y, env_origin_z]
| 8,852 | Python | 50.47093 | 119 | 0.591618 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/utils/usd_utils.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import get_current_stage
from pxr import UsdLux, UsdPhysics
def set_drive_type(prim_path, drive_type):
joint_prim = get_prim_at_path(prim_path)
# set drive type ("angular" or "linear")
drive = UsdPhysics.DriveAPI.Apply(joint_prim, drive_type)
return drive
def set_drive_target_position(drive, target_value):
if not drive.GetTargetPositionAttr():
drive.CreateTargetPositionAttr(target_value)
else:
drive.GetTargetPositionAttr().Set(target_value)
def set_drive_target_velocity(drive, target_value):
if not drive.GetTargetVelocityAttr():
drive.CreateTargetVelocityAttr(target_value)
else:
drive.GetTargetVelocityAttr().Set(target_value)
def set_drive_stiffness(drive, stiffness):
if not drive.GetStiffnessAttr():
drive.CreateStiffnessAttr(stiffness)
else:
drive.GetStiffnessAttr().Set(stiffness)
def set_drive_damping(drive, damping):
if not drive.GetDampingAttr():
drive.CreateDampingAttr(damping)
else:
drive.GetDampingAttr().Set(damping)
def set_drive_max_force(drive, max_force):
if not drive.GetMaxForceAttr():
drive.CreateMaxForceAttr(max_force)
else:
drive.GetMaxForceAttr().Set(max_force)
def set_drive(prim_path, drive_type, target_type, target_value, stiffness, damping, max_force) -> None:
drive = set_drive_type(prim_path, drive_type)
# set target type ("position" or "velocity")
if target_type == "position":
set_drive_target_position(drive, target_value)
elif target_type == "velocity":
set_drive_target_velocity(drive, target_value)
set_drive_stiffness(drive, stiffness)
set_drive_damping(drive, damping)
set_drive_max_force(drive, max_force)
| 3,403 | Python | 36.406593 | 103 | 0.740229 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/shared/reacher.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# Copyright (c) 2022-2023, Johnson Sun
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
# Ref: /omniisaacgymenvs/tasks/shared/reacher.py
import math
from abc import abstractmethod
import numpy as np
import torch
from omni.isaac.core.prims import RigidPrimView, XFormPrim
from omni.isaac.core.scenes.scene import Scene
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import add_reference_to_stage, get_current_stage
from omni.isaac.core.utils.torch import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
# `scale` maps [-1, 1] to [L, U]; `unscale` maps [L, U] to [-1, 1]
from omni.isaac.core.utils.torch import scale, unscale
from omni.isaac.gym.vec_env import VecEnvBase
class ReacherTask(RLTask):
def __init__(
self,
name: str,
env: VecEnvBase,
offset=None
) -> None:
ReacherTask.update_config(self)
RLTask.__init__(self, name, env)
self.x_unit_tensor = torch.tensor([1, 0, 0], dtype=torch.float, device=self.device).repeat((self.num_envs, 1))
self.y_unit_tensor = torch.tensor([0, 1, 0], dtype=torch.float, device=self.device).repeat((self.num_envs, 1))
self.z_unit_tensor = torch.tensor([0, 0, 1], dtype=torch.float, device=self.device).repeat((self.num_envs, 1))
self.reset_goal_buf = self.reset_buf.clone()
self.successes = torch.zeros(self.num_envs, dtype=torch.float, device=self.device)
self.consecutive_successes = torch.zeros(1, dtype=torch.float, device=self.device)
self.av_factor = torch.tensor(self.av_factor, dtype=torch.float, device=self.device)
self.total_successes = 0
self.total_resets = 0
def update_config(self):
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self.dist_reward_scale = self._task_cfg["env"]["distRewardScale"]
self.rot_reward_scale = self._task_cfg["env"]["rotRewardScale"]
self.action_penalty_scale = self._task_cfg["env"]["actionPenaltyScale"]
self.success_tolerance = self._task_cfg["env"]["successTolerance"]
self.reach_goal_bonus = self._task_cfg["env"]["reachGoalBonus"]
self.rot_eps = self._task_cfg["env"]["rotEps"]
self.vel_obs_scale = self._task_cfg["env"]["velObsScale"]
self.reset_position_noise = self._task_cfg["env"]["resetPositionNoise"]
self.reset_rotation_noise = self._task_cfg["env"]["resetRotationNoise"]
self.reset_dof_pos_noise = self._task_cfg["env"]["resetDofPosRandomInterval"]
self.reset_dof_vel_noise = self._task_cfg["env"]["resetDofVelRandomInterval"]
self.arm_dof_speed_scale = self._task_cfg["env"]["dofSpeedScale"]
self.use_relative_control = self._task_cfg["env"]["useRelativeControl"]
self.act_moving_average = self._task_cfg["env"]["actionsMovingAverage"]
self.max_episode_length = self._task_cfg["env"]["episodeLength"]
self.reset_time = self._task_cfg["env"].get("resetTime", -1.0)
self.print_success_stat = self._task_cfg["env"]["printNumSuccesses"]
self.max_consecutive_successes = self._task_cfg["env"]["maxConsecutiveSuccesses"]
self.av_factor = self._task_cfg["env"].get("averFactor", 0.1)
self.dt = 1.0 / 60
control_freq_inv = self._task_cfg["env"].get("controlFrequencyInv", 1)
if self.reset_time > 0.0:
self.max_episode_length = int(round(self.reset_time / (control_freq_inv * self.dt)))
print("Reset time: ", self.reset_time)
print("New episode length: ", self.max_episode_length)
def set_up_scene(self, scene: Scene) -> None:
self._stage = get_current_stage()
self._assets_root_path = 'omniverse://localhost/Projects/J3soon/Isaac/2023.1.0'
self.get_arm()
self.object_start_translation = torch.tensor([0.0, 0.0, 0.0], device=self.device)
self.object_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.goal_displacement_tensor = torch.tensor([0.0, 0.0, 0.0], device=self.device)
self.goal_start_translation = torch.tensor([0.0, 0.0, 0.0], device=self.device) + self.goal_displacement_tensor
self.goal_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.get_object()
self.get_goal()
super().set_up_scene(scene)
self._arms = self.get_arm_view(scene)
scene.add(self._arms)
self._objects = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/object/object",
name="object_view",
reset_xform_properties=False,
)
self._objects._non_root_link = True # hack to ignore kinematics
scene.add(self._objects)
self._goals = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/goal/object", name="goal_view", reset_xform_properties=False
)
self._goals._non_root_link = True # hack to ignore kinematics
scene.add(self._goals)
def initialize_views(self, scene):
RLTask.initialize_views(self, scene)
if scene.object_exists("dofbot_view"):
scene.remove_object("dofbot_view", registry_only=True)
if scene.object_exists("ur10_view"):
scene.remove_object("ur10_view", registry_only=True)
if scene.object_exists("kuka_view"):
scene.remove_object("kuka_view", registry_only=True)
if scene.object_exists("hiwin_view"):
scene.remove_object("hiwin_view", registry_only=True)
if scene.object_exists("goal_view"):
scene.remove_object("goal_view", registry_only=True)
if scene.object_exists("object_view"):
scene.remove_object("object_view", registry_only=True)
self.object_start_translation = torch.tensor([0.0, 0.0, 0.0], device=self.device)
self.object_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.goal_displacement_tensor = torch.tensor([0.0, 0.0, 0.0], device=self.device)
self.goal_start_translation = torch.tensor([0.0, 0.0, 0.0], device=self.device) + self.goal_displacement_tensor
self.goal_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self._arms = self.get_arm_view(scene)
scene.add(self._arms)
self._objects = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/object/object",
name="object_view",
reset_xform_properties=False,
)
self._objects._non_root_link = True # hack to ignore kinematics
scene.add(self._objects)
self._goals = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/goal/object", name="goal_view", reset_xform_properties=False
)
self._goals._non_root_link = True # hack to ignore kinematics
scene.add(self._goals)
@abstractmethod
def get_num_dof(self):
pass
@abstractmethod
def get_arm(self):
pass
@abstractmethod
def get_arm_view(self):
pass
@abstractmethod
def get_observations(self):
pass
@abstractmethod
def get_reset_target_new_pos(self, n_reset_envs):
pass
@abstractmethod
def send_joint_pos(self, joint_pos):
pass
def get_object(self):
self.object_usd_path = f"{self._assets_root_path}/Isaac/Props/Blocks/block_instanceable.usd"
add_reference_to_stage(self.object_usd_path, self.default_zero_env_path + "/object")
obj = XFormPrim(
prim_path=self.default_zero_env_path + "/object/object",
name="object",
translation=self.object_start_translation,
orientation=self.object_start_orientation,
scale=self.object_scale,
)
self._sim_config.apply_articulation_settings(
"object", get_prim_at_path(obj.prim_path), self._sim_config.parse_actor_config("object")
)
def get_goal(self):
self.goal_usd_path = f"{self._assets_root_path}/Isaac/Props/Blocks/block_instanceable.usd"
add_reference_to_stage(self.goal_usd_path, self.default_zero_env_path + "/goal")
goal = XFormPrim(
prim_path=self.default_zero_env_path + "/goal/object",
name="goal",
translation=self.goal_start_translation,
orientation=self.goal_start_orientation,
scale=self.goal_scale
)
self._sim_config.apply_articulation_settings("goal", get_prim_at_path(goal.prim_path), self._sim_config.parse_actor_config("goal_object"))
def post_reset(self):
self.num_arm_dofs = self.get_num_dof()
self.actuated_dof_indices = torch.arange(self.num_arm_dofs, dtype=torch.long, device=self.device)
self.arm_dof_targets = torch.zeros((self.num_envs, self._arms.num_dof), dtype=torch.float, device=self.device)
self.prev_targets = torch.zeros((self.num_envs, self.num_arm_dofs), dtype=torch.float, device=self.device)
self.cur_targets = torch.zeros((self.num_envs, self.num_arm_dofs), dtype=torch.float, device=self.device)
dof_limits = self._dof_limits[:, :self.num_arm_dofs]
self.arm_dof_lower_limits, self.arm_dof_upper_limits = torch.t(dof_limits[0].to(self.device))
self.arm_dof_default_pos = torch.zeros(self.num_arm_dofs, dtype=torch.float, device=self.device)
self.arm_dof_default_vel = torch.zeros(self.num_arm_dofs, dtype=torch.float, device=self.device)
self.end_effectors_init_pos, self.end_effectors_init_rot = self._arms._end_effectors.get_world_poses()
self.goal_pos, self.goal_rot = self._goals.get_world_poses()
self.goal_pos -= self._env_pos
# randomize all envs
indices = torch.arange(self._num_envs, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
def calculate_metrics(self):
self.fall_dist = 0
self.fall_penalty = 0
(
self.rew_buf[:],
self.reset_buf[:],
self.reset_goal_buf[:],
self.progress_buf[:],
self.successes[:],
self.consecutive_successes[:],
) = compute_arm_reward(
self.rew_buf,
self.reset_buf,
self.reset_goal_buf,
self.progress_buf,
self.successes,
self.consecutive_successes,
self.max_episode_length,
self.object_pos,
self.object_rot,
self.goal_pos,
self.goal_rot,
self.dist_reward_scale,
self.rot_reward_scale,
self.rot_eps,
self.actions,
self.action_penalty_scale,
self.success_tolerance,
self.reach_goal_bonus,
self.fall_dist,
self.fall_penalty,
self.max_consecutive_successes,
self.av_factor,
)
self.extras["consecutive_successes"] = self.consecutive_successes.mean()
if self.print_success_stat:
self.total_resets = self.total_resets + self.reset_buf.sum()
direct_average_successes = self.total_successes + self.successes.sum()
self.total_successes = self.total_successes + (self.successes * self.reset_buf).sum()
# The direct average shows the overall result more quickly, but slightly undershoots long term policy performance.
print(
"Direct average consecutive successes = {:.1f}".format(
direct_average_successes / (self.total_resets + self.num_envs)
)
)
if self.total_resets > 0:
print(
"Post-Reset average consecutive successes = {:.1f}".format(self.total_successes / self.total_resets)
)
def pre_physics_step(self, actions):
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
goal_env_ids = self.reset_goal_buf.nonzero(as_tuple=False).squeeze(-1)
end_effectors_pos, end_effectors_rot = self._arms._end_effectors.get_world_poses()
# Reverse the default rotation and rotate the displacement tensor according to the current rotation
self.object_pos = end_effectors_pos + quat_rotate(end_effectors_rot, quat_rotate_inverse(self.end_effectors_init_rot, self.get_object_displacement_tensor()))
self.object_pos -= self._env_pos # subtract world env pos
self.object_rot = end_effectors_rot
object_pos = self.object_pos + self._env_pos
object_rot = self.object_rot
self._objects.set_world_poses(object_pos, object_rot)
# if only goals need reset, then call set API
if len(goal_env_ids) > 0 and len(env_ids) == 0:
self.reset_target_pose(goal_env_ids)
elif len(goal_env_ids) > 0:
self.reset_target_pose(goal_env_ids)
if len(env_ids) > 0:
self.reset_idx(env_ids)
self.actions = actions.clone().to(self.device)
# Reacher tasks don't require gripper actions, disable it.
self.actions[:, 5] = 0.0
if self.use_relative_control:
targets = (
self.prev_targets[:, self.actuated_dof_indices] + self.arm_dof_speed_scale * self.dt * self.actions
)
self.cur_targets[:, self.actuated_dof_indices] = tensor_clamp(
targets,
self.arm_dof_lower_limits[self.actuated_dof_indices],
self.arm_dof_upper_limits[self.actuated_dof_indices],
)
else:
self.cur_targets[:, self.actuated_dof_indices] = scale(
self.actions[:, :self.num_arm_dofs],
self.arm_dof_lower_limits[self.actuated_dof_indices],
self.arm_dof_upper_limits[self.actuated_dof_indices],
)
self.cur_targets[:, self.actuated_dof_indices] = (
self.act_moving_average * self.cur_targets[:, self.actuated_dof_indices]
+ (1.0 - self.act_moving_average) * self.prev_targets[:, self.actuated_dof_indices]
)
self.cur_targets[:, self.actuated_dof_indices] = tensor_clamp(
self.cur_targets[:, self.actuated_dof_indices],
self.arm_dof_lower_limits[self.actuated_dof_indices],
self.arm_dof_upper_limits[self.actuated_dof_indices],
)
self.prev_targets[:, self.actuated_dof_indices] = self.cur_targets[:, self.actuated_dof_indices]
self._arms.set_joint_position_targets(
self.cur_targets[:, self.actuated_dof_indices], indices=None, joint_indices=self.actuated_dof_indices
)
if self._task_cfg['sim2real']['enabled'] and self.test and self.num_envs == 1:
# Only retrieve the 0-th joint position even when multiple envs are used
cur_joint_pos = self._arms.get_joint_positions(indices=[0], joint_indices=self.actuated_dof_indices)
# Send the current joint positions to the real robot
joint_pos = cur_joint_pos[0]
if torch.any(joint_pos < self.arm_dof_lower_limits) or torch.any(joint_pos > self.arm_dof_upper_limits):
print("get_joint_positions out of bound, send_joint_pos skipped")
else:
self.send_joint_pos(joint_pos)
def is_done(self):
pass
def reset_target_pose(self, env_ids):
# reset goal
indices = env_ids.to(dtype=torch.int32)
rand_floats = torch_rand_float(-1.0, 1.0, (len(env_ids), 4), device=self.device)
new_pos = self.get_reset_target_new_pos(len(env_ids))
new_rot = randomize_rotation(
rand_floats[:, 0], rand_floats[:, 1], self.x_unit_tensor[env_ids], self.y_unit_tensor[env_ids]
)
self.goal_pos[env_ids] = new_pos
self.goal_rot[env_ids] = new_rot
goal_pos, goal_rot = self.goal_pos.clone(), self.goal_rot.clone()
goal_pos[env_ids] = (
self.goal_pos[env_ids] + self._env_pos[env_ids]
) # add world env pos
self._goals.set_world_poses(goal_pos[env_ids], goal_rot[env_ids], indices)
self.reset_goal_buf[env_ids] = 0
def reset_idx(self, env_ids):
indices = env_ids.to(dtype=torch.int32)
rand_floats = torch_rand_float(-1.0, 1.0, (len(env_ids), self.num_arm_dofs * 2 + 5), device=self.device)
self.reset_target_pose(env_ids)
# reset arm
delta_max = self.arm_dof_upper_limits - self.arm_dof_default_pos
delta_min = self.arm_dof_lower_limits - self.arm_dof_default_pos
rand_delta = delta_min + (delta_max - delta_min) * (rand_floats[:, 5:5+self.num_arm_dofs] + 1.0) * 0.5
pos = self.arm_dof_default_pos + self.reset_dof_pos_noise * rand_delta
dof_pos = torch.zeros((self.num_envs, self._arms.num_dof), device=self.device)
dof_pos[env_ids, :self.num_arm_dofs] = pos
dof_vel = torch.zeros((self.num_envs, self._arms.num_dof), device=self.device)
dof_vel[env_ids, :self.num_arm_dofs] = self.arm_dof_default_vel + \
self.reset_dof_vel_noise * rand_floats[:, 5+self.num_arm_dofs:5+self.num_arm_dofs*2]
self.prev_targets[env_ids, :self.num_arm_dofs] = pos
self.cur_targets[env_ids, :self.num_arm_dofs] = pos
self.arm_dof_targets[env_ids, :self.num_arm_dofs] = pos
self._arms.set_joint_position_targets(self.arm_dof_targets[env_ids], indices)
# set_joint_positions doesn't seem to apply immediately.
self._arms.set_joint_positions(dof_pos[env_ids], indices)
self._arms.set_joint_velocities(dof_vel[env_ids], indices)
self.progress_buf[env_ids] = 0
self.reset_buf[env_ids] = 0
self.successes[env_ids] = 0
#####################################################################
###=========================jit functions=========================###
#####################################################################
@torch.jit.script
def randomize_rotation(rand0, rand1, x_unit_tensor, y_unit_tensor):
return quat_mul(
quat_from_angle_axis(rand0 * np.pi, x_unit_tensor), quat_from_angle_axis(rand1 * np.pi, y_unit_tensor)
)
@torch.jit.script
def compute_arm_reward(
rew_buf,
reset_buf,
reset_goal_buf,
progress_buf,
successes,
consecutive_successes,
max_episode_length: float,
object_pos,
object_rot,
target_pos,
target_rot,
dist_reward_scale: float,
rot_reward_scale: float,
rot_eps: float,
actions,
action_penalty_scale: float,
success_tolerance: float,
reach_goal_bonus: float,
fall_dist: float,
fall_penalty: float,
max_consecutive_successes: int,
av_factor: float,
):
goal_dist = torch.norm(object_pos - target_pos, p=2, dim=-1)
# Orientation alignment for the cube in hand and goal cube
quat_diff = quat_mul(object_rot, quat_conjugate(target_rot))
rot_dist = 2.0 * torch.asin(
torch.clamp(torch.norm(quat_diff[:, 1:4], p=2, dim=-1), max=1.0)
) # changed quat convention
dist_rew = goal_dist * dist_reward_scale
rot_rew = 1.0 / (torch.abs(rot_dist) + rot_eps) * rot_reward_scale
action_penalty = torch.sum(actions**2, dim=-1)
# Total reward is: position distance + orientation alignment + action regularization + success bonus + fall penalty
reward = dist_rew + action_penalty * action_penalty_scale
# Find out which envs hit the goal and update successes count
goal_resets = torch.where(torch.abs(goal_dist) <= success_tolerance, torch.ones_like(reset_goal_buf), reset_goal_buf)
successes = successes + goal_resets
# Success bonus: orientation is within `success_tolerance` of goal orientation
reward = torch.where(goal_resets == 1, reward + reach_goal_bonus, reward)
resets = reset_buf
if max_consecutive_successes > 0:
# Reset progress buffer on goal envs if max_consecutive_successes > 0
progress_buf = torch.where(
torch.abs(rot_dist) <= success_tolerance, torch.zeros_like(progress_buf), progress_buf
)
resets = torch.where(successes >= max_consecutive_successes, torch.ones_like(resets), resets)
resets = torch.where(progress_buf >= max_episode_length - 1, torch.ones_like(resets), resets)
num_resets = torch.sum(resets)
finished_cons_successes = torch.sum(successes * resets.float())
cons_successes = torch.where(
num_resets > 0,
av_factor * finished_cons_successes / num_resets + (1.0 - av_factor) * consecutive_successes,
consecutive_successes,
)
return reward, resets, goal_resets, progress_buf, successes, cons_successes
| 22,312 | Python | 42.836935 | 165 | 0.629482 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/shared/in_hand_manipulation.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
from abc import abstractmethod
import numpy as np
import torch
from omni.isaac.core.prims import RigidPrimView, XFormPrim
from omni.isaac.core.utils.nucleus import get_assets_root_path
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.stage import add_reference_to_stage, get_current_stage
from omni.isaac.core.utils.torch import *
from omniisaacgymenvs.tasks.base.rl_task import RLTask
class InHandManipulationTask(RLTask):
def __init__(self, name, env, offset=None) -> None:
InHandManipulationTask.update_config(self)
RLTask.__init__(self, name, env)
self.x_unit_tensor = torch.tensor([1, 0, 0], dtype=torch.float, device=self.device).repeat((self.num_envs, 1))
self.y_unit_tensor = torch.tensor([0, 1, 0], dtype=torch.float, device=self.device).repeat((self.num_envs, 1))
self.z_unit_tensor = torch.tensor([0, 0, 1], dtype=torch.float, device=self.device).repeat((self.num_envs, 1))
self.reset_goal_buf = self.reset_buf.clone()
self.successes = torch.zeros(self.num_envs, dtype=torch.float, device=self.device)
self.consecutive_successes = torch.zeros(1, dtype=torch.float, device=self.device)
self.randomization_buf = torch.zeros(self.num_envs, dtype=torch.long, device=self.device)
self.av_factor = torch.tensor(self.av_factor, dtype=torch.float, device=self.device)
self.total_successes = 0
self.total_resets = 0
def update_config(self):
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self.dist_reward_scale = self._task_cfg["env"]["distRewardScale"]
self.rot_reward_scale = self._task_cfg["env"]["rotRewardScale"]
self.action_penalty_scale = self._task_cfg["env"]["actionPenaltyScale"]
self.success_tolerance = self._task_cfg["env"]["successTolerance"]
self.reach_goal_bonus = self._task_cfg["env"]["reachGoalBonus"]
self.fall_dist = self._task_cfg["env"]["fallDistance"]
self.fall_penalty = self._task_cfg["env"]["fallPenalty"]
self.rot_eps = self._task_cfg["env"]["rotEps"]
self.vel_obs_scale = self._task_cfg["env"]["velObsScale"]
self.reset_position_noise = self._task_cfg["env"]["resetPositionNoise"]
self.reset_rotation_noise = self._task_cfg["env"]["resetRotationNoise"]
self.reset_dof_pos_noise = self._task_cfg["env"]["resetDofPosRandomInterval"]
self.reset_dof_vel_noise = self._task_cfg["env"]["resetDofVelRandomInterval"]
self.hand_dof_speed_scale = self._task_cfg["env"]["dofSpeedScale"]
self.use_relative_control = self._task_cfg["env"]["useRelativeControl"]
self.act_moving_average = self._task_cfg["env"]["actionsMovingAverage"]
self.max_episode_length = self._task_cfg["env"]["episodeLength"]
self.reset_time = self._task_cfg["env"].get("resetTime", -1.0)
self.print_success_stat = self._task_cfg["env"]["printNumSuccesses"]
self.max_consecutive_successes = self._task_cfg["env"]["maxConsecutiveSuccesses"]
self.av_factor = self._task_cfg["env"].get("averFactor", 0.1)
self.dt = 1.0 / 60
control_freq_inv = self._task_cfg["env"].get("controlFrequencyInv", 1)
if self.reset_time > 0.0:
self.max_episode_length = int(round(self.reset_time / (control_freq_inv * self.dt)))
print("Reset time: ", self.reset_time)
print("New episode length: ", self.max_episode_length)
def set_up_scene(self, scene) -> None:
self._stage = get_current_stage()
self._assets_root_path = get_assets_root_path()
self.get_starting_positions()
self.get_hand()
self.object_start_translation = self.hand_start_translation.clone()
self.object_start_translation[1] += self.pose_dy
self.object_start_translation[2] += self.pose_dz
self.object_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.goal_displacement_tensor = torch.tensor([-0.2, -0.06, 0.12], device=self.device)
self.goal_start_translation = self.object_start_translation + self.goal_displacement_tensor
self.goal_start_translation[2] -= 0.04
self.goal_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.get_object(self.hand_start_translation, self.pose_dy, self.pose_dz)
self.get_goal()
super().set_up_scene(scene, filter_collisions=False)
self._hands = self.get_hand_view(scene)
scene.add(self._hands)
self._objects = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/object/object",
name="object_view",
reset_xform_properties=False,
masses=torch.tensor([0.07087] * self._num_envs, device=self.device),
)
scene.add(self._objects)
self._goals = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/goal/object", name="goal_view", reset_xform_properties=False
)
self._goals._non_root_link = True # hack to ignore kinematics
scene.add(self._goals)
if self._dr_randomizer.randomize:
self._dr_randomizer.apply_on_startup_domain_randomization(self)
def initialize_views(self, scene):
RLTask.initialize_views(self, scene)
if scene.object_exists("shadow_hand_view"):
scene.remove_object("shadow_hand_view", registry_only=True)
if scene.object_exists("finger_view"):
scene.remove_object("finger_view", registry_only=True)
if scene.object_exists("allegro_hand_view"):
scene.remove_object("allegro_hand_view", registry_only=True)
if scene.object_exists("goal_view"):
scene.remove_object("goal_view", registry_only=True)
if scene.object_exists("object_view"):
scene.remove_object("object_view", registry_only=True)
self.get_starting_positions()
self.object_start_translation = self.hand_start_translation.clone()
self.object_start_translation[1] += self.pose_dy
self.object_start_translation[2] += self.pose_dz
self.object_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self.goal_displacement_tensor = torch.tensor([-0.2, -0.06, 0.12], device=self.device)
self.goal_start_translation = self.object_start_translation + self.goal_displacement_tensor
self.goal_start_translation[2] -= 0.04
self.goal_start_orientation = torch.tensor([1.0, 0.0, 0.0, 0.0], device=self.device)
self._hands = self.get_hand_view(scene)
scene.add(self._hands)
self._objects = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/object/object",
name="object_view",
reset_xform_properties=False,
masses=torch.tensor([0.07087] * self._num_envs, device=self.device),
)
scene.add(self._objects)
self._goals = RigidPrimView(
prim_paths_expr="/World/envs/env_.*/goal/object", name="goal_view", reset_xform_properties=False
)
self._goals._non_root_link = True # hack to ignore kinematics
scene.add(self._goals)
if self._dr_randomizer.randomize:
self._dr_randomizer.apply_on_startup_domain_randomization(self)
@abstractmethod
def get_hand(self):
pass
@abstractmethod
def get_hand_view(self):
pass
@abstractmethod
def get_observations(self):
pass
def get_object(self, hand_start_translation, pose_dy, pose_dz):
self.object_usd_path = f"{self._assets_root_path}/Isaac/Props/Blocks/block_instanceable.usd"
add_reference_to_stage(self.object_usd_path, self.default_zero_env_path + "/object")
obj = XFormPrim(
prim_path=self.default_zero_env_path + "/object/object",
name="object",
translation=self.object_start_translation,
orientation=self.object_start_orientation,
scale=self.object_scale,
)
self._sim_config.apply_articulation_settings(
"object", get_prim_at_path(obj.prim_path), self._sim_config.parse_actor_config("object")
)
def get_goal(self):
add_reference_to_stage(self.object_usd_path, self.default_zero_env_path + "/goal")
goal = XFormPrim(
prim_path=self.default_zero_env_path + "/goal",
name="goal",
translation=self.goal_start_translation,
orientation=self.goal_start_orientation,
scale=self.object_scale,
)
self._sim_config.apply_articulation_settings(
"goal", get_prim_at_path(goal.prim_path), self._sim_config.parse_actor_config("goal_object")
)
def post_reset(self):
self.num_hand_dofs = self._hands.num_dof
self.actuated_dof_indices = self._hands.actuated_dof_indices
self.hand_dof_targets = torch.zeros((self.num_envs, self.num_hand_dofs), dtype=torch.float, device=self.device)
self.prev_targets = torch.zeros((self.num_envs, self.num_hand_dofs), dtype=torch.float, device=self.device)
self.cur_targets = torch.zeros((self.num_envs, self.num_hand_dofs), dtype=torch.float, device=self.device)
dof_limits = self._hands.get_dof_limits()
self.hand_dof_lower_limits, self.hand_dof_upper_limits = torch.t(dof_limits[0].to(self.device))
self.hand_dof_default_pos = torch.zeros(self.num_hand_dofs, dtype=torch.float, device=self.device)
self.hand_dof_default_vel = torch.zeros(self.num_hand_dofs, dtype=torch.float, device=self.device)
self.object_init_pos, self.object_init_rot = self._objects.get_world_poses()
self.object_init_pos -= self._env_pos
self.object_init_velocities = torch.zeros_like(
self._objects.get_velocities(), dtype=torch.float, device=self.device
)
self.goal_pos = self.object_init_pos.clone()
self.goal_pos[:, 2] -= 0.04
self.goal_rot = self.object_init_rot.clone()
self.goal_init_pos = self.goal_pos.clone()
self.goal_init_rot = self.goal_rot.clone()
# randomize all envs
indices = torch.arange(self._num_envs, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
if self._dr_randomizer.randomize:
self._dr_randomizer.set_up_domain_randomization(self)
def get_object_goal_observations(self):
self.object_pos, self.object_rot = self._objects.get_world_poses(clone=False)
self.object_pos -= self._env_pos
self.object_velocities = self._objects.get_velocities(clone=False)
self.object_linvel = self.object_velocities[:, 0:3]
self.object_angvel = self.object_velocities[:, 3:6]
def calculate_metrics(self):
(
self.rew_buf[:],
self.reset_buf[:],
self.reset_goal_buf[:],
self.progress_buf[:],
self.successes[:],
self.consecutive_successes[:],
) = compute_hand_reward(
self.rew_buf,
self.reset_buf,
self.reset_goal_buf,
self.progress_buf,
self.successes,
self.consecutive_successes,
self.max_episode_length,
self.object_pos,
self.object_rot,
self.goal_pos,
self.goal_rot,
self.dist_reward_scale,
self.rot_reward_scale,
self.rot_eps,
self.actions,
self.action_penalty_scale,
self.success_tolerance,
self.reach_goal_bonus,
self.fall_dist,
self.fall_penalty,
self.max_consecutive_successes,
self.av_factor,
)
self.extras["consecutive_successes"] = self.consecutive_successes.mean()
self.randomization_buf += 1
if self.print_success_stat:
self.total_resets = self.total_resets + self.reset_buf.sum()
direct_average_successes = self.total_successes + self.successes.sum()
self.total_successes = self.total_successes + (self.successes * self.reset_buf).sum()
# The direct average shows the overall result more quickly, but slightly undershoots long term policy performance.
print(
"Direct average consecutive successes = {:.1f}".format(
direct_average_successes / (self.total_resets + self.num_envs)
)
)
if self.total_resets > 0:
print(
"Post-Reset average consecutive successes = {:.1f}".format(self.total_successes / self.total_resets)
)
def pre_physics_step(self, actions):
if not self._env._world.is_playing():
return
env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
goal_env_ids = self.reset_goal_buf.nonzero(as_tuple=False).squeeze(-1)
reset_buf = self.reset_buf.clone()
# if only goals need reset, then call set API
if len(goal_env_ids) > 0 and len(env_ids) == 0:
self.reset_target_pose(goal_env_ids)
elif len(goal_env_ids) > 0:
self.reset_target_pose(goal_env_ids)
if len(env_ids) > 0:
self.reset_idx(env_ids)
self.actions = actions.clone().to(self.device)
if self.use_relative_control:
targets = (
self.prev_targets[:, self.actuated_dof_indices] + self.hand_dof_speed_scale * self.dt * self.actions
)
self.cur_targets[:, self.actuated_dof_indices] = tensor_clamp(
targets,
self.hand_dof_lower_limits[self.actuated_dof_indices],
self.hand_dof_upper_limits[self.actuated_dof_indices],
)
else:
self.cur_targets[:, self.actuated_dof_indices] = scale(
self.actions,
self.hand_dof_lower_limits[self.actuated_dof_indices],
self.hand_dof_upper_limits[self.actuated_dof_indices],
)
self.cur_targets[:, self.actuated_dof_indices] = (
self.act_moving_average * self.cur_targets[:, self.actuated_dof_indices]
+ (1.0 - self.act_moving_average) * self.prev_targets[:, self.actuated_dof_indices]
)
self.cur_targets[:, self.actuated_dof_indices] = tensor_clamp(
self.cur_targets[:, self.actuated_dof_indices],
self.hand_dof_lower_limits[self.actuated_dof_indices],
self.hand_dof_upper_limits[self.actuated_dof_indices],
)
self.prev_targets[:, self.actuated_dof_indices] = self.cur_targets[:, self.actuated_dof_indices]
self._hands.set_joint_position_targets(
self.cur_targets[:, self.actuated_dof_indices], indices=None, joint_indices=self.actuated_dof_indices
)
if self._dr_randomizer.randomize:
rand_envs = torch.where(
self.randomization_buf >= self._dr_randomizer.min_frequency,
torch.ones_like(self.randomization_buf),
torch.zeros_like(self.randomization_buf),
)
rand_env_ids = torch.nonzero(torch.logical_and(rand_envs, reset_buf))
self.dr.physics_view.step_randomization(rand_env_ids)
self.randomization_buf[rand_env_ids] = 0
def is_done(self):
pass
def reset_target_pose(self, env_ids):
# reset goal
indices = env_ids.to(dtype=torch.int32)
rand_floats = torch_rand_float(-1.0, 1.0, (len(env_ids), 4), device=self.device)
new_rot = randomize_rotation(
rand_floats[:, 0], rand_floats[:, 1], self.x_unit_tensor[env_ids], self.y_unit_tensor[env_ids]
)
self.goal_pos[env_ids] = self.goal_init_pos[env_ids, 0:3]
self.goal_rot[env_ids] = new_rot
goal_pos, goal_rot = self.goal_pos.clone(), self.goal_rot.clone()
goal_pos[env_ids] = (
self.goal_pos[env_ids] + self.goal_displacement_tensor + self._env_pos[env_ids]
) # add world env pos
self._goals.set_world_poses(goal_pos[env_ids], goal_rot[env_ids], indices)
self.reset_goal_buf[env_ids] = 0
def reset_idx(self, env_ids):
indices = env_ids.to(dtype=torch.int32)
rand_floats = torch_rand_float(-1.0, 1.0, (len(env_ids), self.num_hand_dofs * 2 + 5), device=self.device)
self.reset_target_pose(env_ids)
# reset object
new_object_pos = (
self.object_init_pos[env_ids] + self.reset_position_noise * rand_floats[:, 0:3] + self._env_pos[env_ids]
) # add world env pos
new_object_rot = randomize_rotation(
rand_floats[:, 3], rand_floats[:, 4], self.x_unit_tensor[env_ids], self.y_unit_tensor[env_ids]
)
object_velocities = torch.zeros_like(self.object_init_velocities, dtype=torch.float, device=self.device)
self._objects.set_velocities(object_velocities[env_ids], indices)
self._objects.set_world_poses(new_object_pos, new_object_rot, indices)
# reset hand
delta_max = self.hand_dof_upper_limits - self.hand_dof_default_pos
delta_min = self.hand_dof_lower_limits - self.hand_dof_default_pos
rand_delta = delta_min + (delta_max - delta_min) * 0.5 * (rand_floats[:, 5 : 5 + self.num_hand_dofs] + 1.0)
pos = self.hand_dof_default_pos + self.reset_dof_pos_noise * rand_delta
dof_pos = torch.zeros((self.num_envs, self.num_hand_dofs), device=self.device)
dof_pos[env_ids, :] = pos
dof_vel = torch.zeros((self.num_envs, self.num_hand_dofs), device=self.device)
dof_vel[env_ids, :] = (
self.hand_dof_default_vel
+ self.reset_dof_vel_noise * rand_floats[:, 5 + self.num_hand_dofs : 5 + self.num_hand_dofs * 2]
)
self.prev_targets[env_ids, : self.num_hand_dofs] = pos
self.cur_targets[env_ids, : self.num_hand_dofs] = pos
self.hand_dof_targets[env_ids, :] = pos
self._hands.set_joint_position_targets(self.hand_dof_targets[env_ids], indices)
self._hands.set_joint_positions(dof_pos[env_ids], indices)
self._hands.set_joint_velocities(dof_vel[env_ids], indices)
self.progress_buf[env_ids] = 0
self.reset_buf[env_ids] = 0
self.successes[env_ids] = 0
#####################################################################
###=========================jit functions=========================###
#####################################################################
@torch.jit.script
def randomize_rotation(rand0, rand1, x_unit_tensor, y_unit_tensor):
return quat_mul(
quat_from_angle_axis(rand0 * np.pi, x_unit_tensor), quat_from_angle_axis(rand1 * np.pi, y_unit_tensor)
)
@torch.jit.script
def compute_hand_reward(
rew_buf,
reset_buf,
reset_goal_buf,
progress_buf,
successes,
consecutive_successes,
max_episode_length: float,
object_pos,
object_rot,
target_pos,
target_rot,
dist_reward_scale: float,
rot_reward_scale: float,
rot_eps: float,
actions,
action_penalty_scale: float,
success_tolerance: float,
reach_goal_bonus: float,
fall_dist: float,
fall_penalty: float,
max_consecutive_successes: int,
av_factor: float,
):
goal_dist = torch.norm(object_pos - target_pos, p=2, dim=-1)
# Orientation alignment for the cube in hand and goal cube
quat_diff = quat_mul(object_rot, quat_conjugate(target_rot))
rot_dist = 2.0 * torch.asin(
torch.clamp(torch.norm(quat_diff[:, 1:4], p=2, dim=-1), max=1.0)
) # changed quat convention
dist_rew = goal_dist * dist_reward_scale
rot_rew = 1.0 / (torch.abs(rot_dist) + rot_eps) * rot_reward_scale
action_penalty = torch.sum(actions**2, dim=-1)
# Total reward is: position distance + orientation alignment + action regularization + success bonus + fall penalty
reward = dist_rew + rot_rew + action_penalty * action_penalty_scale
# Find out which envs hit the goal and update successes count
goal_resets = torch.where(torch.abs(rot_dist) <= success_tolerance, torch.ones_like(reset_goal_buf), reset_goal_buf)
successes = successes + goal_resets
# Success bonus: orientation is within `success_tolerance` of goal orientation
reward = torch.where(goal_resets == 1, reward + reach_goal_bonus, reward)
# Fall penalty: distance to the goal is larger than a threashold
reward = torch.where(goal_dist >= fall_dist, reward + fall_penalty, reward)
# Check env termination conditions, including maximum success number
resets = torch.where(goal_dist >= fall_dist, torch.ones_like(reset_buf), reset_buf)
if max_consecutive_successes > 0:
# Reset progress buffer on goal envs if max_consecutive_successes > 0
progress_buf = torch.where(
torch.abs(rot_dist) <= success_tolerance, torch.zeros_like(progress_buf), progress_buf
)
resets = torch.where(successes >= max_consecutive_successes, torch.ones_like(resets), resets)
resets = torch.where(progress_buf >= max_episode_length - 1, torch.ones_like(resets), resets)
# Apply penalty for not reaching the goal
if max_consecutive_successes > 0:
reward = torch.where(progress_buf >= max_episode_length - 1, reward + 0.5 * fall_penalty, reward)
num_resets = torch.sum(resets)
finished_cons_successes = torch.sum(successes * resets.float())
cons_successes = torch.where(
num_resets > 0,
av_factor * finished_cons_successes / num_resets + (1.0 - av_factor) * consecutive_successes,
consecutive_successes,
)
return reward, resets, goal_resets, progress_buf, successes, cons_successes
| 23,472 | Python | 43.12218 | 126 | 0.630624 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/tasks/shared/locomotion.py | # Copyright (c) 2018-2022, NVIDIA Corporation
# All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
# list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
import math
from abc import abstractmethod
import numpy as np
import torch
from omni.isaac.core.articulations import ArticulationView
from omni.isaac.core.utils.prims import get_prim_at_path
from omni.isaac.core.utils.torch.maths import tensor_clamp, torch_rand_float, unscale
from omni.isaac.core.utils.torch.rotations import compute_heading_and_up, compute_rot, quat_conjugate
from omniisaacgymenvs.tasks.base.rl_task import RLTask
class LocomotionTask(RLTask):
def __init__(self, name, env, offset=None) -> None:
LocomotionTask.update_config(self)
RLTask.__init__(self, name, env)
return
def update_config(self):
self._num_envs = self._task_cfg["env"]["numEnvs"]
self._env_spacing = self._task_cfg["env"]["envSpacing"]
self._max_episode_length = self._task_cfg["env"]["episodeLength"]
self.dof_vel_scale = self._task_cfg["env"]["dofVelocityScale"]
self.angular_velocity_scale = self._task_cfg["env"]["angularVelocityScale"]
self.contact_force_scale = self._task_cfg["env"]["contactForceScale"]
self.power_scale = self._task_cfg["env"]["powerScale"]
self.heading_weight = self._task_cfg["env"]["headingWeight"]
self.up_weight = self._task_cfg["env"]["upWeight"]
self.actions_cost_scale = self._task_cfg["env"]["actionsCost"]
self.energy_cost_scale = self._task_cfg["env"]["energyCost"]
self.joints_at_limit_cost_scale = self._task_cfg["env"]["jointsAtLimitCost"]
self.death_cost = self._task_cfg["env"]["deathCost"]
self.termination_height = self._task_cfg["env"]["terminationHeight"]
self.alive_reward_scale = self._task_cfg["env"]["alive_reward_scale"]
@abstractmethod
def set_up_scene(self, scene) -> None:
pass
@abstractmethod
def get_robot(self):
pass
def get_observations(self) -> dict:
torso_position, torso_rotation = self._robots.get_world_poses(clone=False)
velocities = self._robots.get_velocities(clone=False)
velocity = velocities[:, 0:3]
ang_velocity = velocities[:, 3:6]
dof_pos = self._robots.get_joint_positions(clone=False)
dof_vel = self._robots.get_joint_velocities(clone=False)
# force sensors attached to the feet
sensor_force_torques = self._robots.get_measured_joint_forces(joint_indices=self._sensor_indices)
(
self.obs_buf[:],
self.potentials[:],
self.prev_potentials[:],
self.up_vec[:],
self.heading_vec[:],
) = get_observations(
torso_position,
torso_rotation,
velocity,
ang_velocity,
dof_pos,
dof_vel,
self.targets,
self.potentials,
self.dt,
self.inv_start_rot,
self.basis_vec0,
self.basis_vec1,
self.dof_limits_lower,
self.dof_limits_upper,
self.dof_vel_scale,
sensor_force_torques,
self._num_envs,
self.contact_force_scale,
self.actions,
self.angular_velocity_scale,
)
observations = {self._robots.name: {"obs_buf": self.obs_buf}}
return observations
def pre_physics_step(self, actions) -> None:
if not self._env._world.is_playing():
return
reset_env_ids = self.reset_buf.nonzero(as_tuple=False).squeeze(-1)
if len(reset_env_ids) > 0:
self.reset_idx(reset_env_ids)
self.actions = actions.clone().to(self._device)
forces = self.actions * self.joint_gears * self.power_scale
indices = torch.arange(self._robots.count, dtype=torch.int32, device=self._device)
# applies joint torques
self._robots.set_joint_efforts(forces, indices=indices)
def reset_idx(self, env_ids):
num_resets = len(env_ids)
# randomize DOF positions and velocities
dof_pos = torch_rand_float(-0.2, 0.2, (num_resets, self._robots.num_dof), device=self._device)
dof_pos[:] = tensor_clamp(self.initial_dof_pos[env_ids] + dof_pos, self.dof_limits_lower, self.dof_limits_upper)
dof_vel = torch_rand_float(-0.1, 0.1, (num_resets, self._robots.num_dof), device=self._device)
root_pos, root_rot = self.initial_root_pos[env_ids], self.initial_root_rot[env_ids]
root_vel = torch.zeros((num_resets, 6), device=self._device)
# apply resets
self._robots.set_joint_positions(dof_pos, indices=env_ids)
self._robots.set_joint_velocities(dof_vel, indices=env_ids)
self._robots.set_world_poses(root_pos, root_rot, indices=env_ids)
self._robots.set_velocities(root_vel, indices=env_ids)
to_target = self.targets[env_ids] - self.initial_root_pos[env_ids]
to_target[:, 2] = 0.0
self.prev_potentials[env_ids] = -torch.norm(to_target, p=2, dim=-1) / self.dt
self.potentials[env_ids] = self.prev_potentials[env_ids].clone()
# bookkeeping
self.reset_buf[env_ids] = 0
self.progress_buf[env_ids] = 0
num_resets = len(env_ids)
def post_reset(self):
self._robots = self.get_robot()
self.initial_root_pos, self.initial_root_rot = self._robots.get_world_poses()
self.initial_dof_pos = self._robots.get_joint_positions()
# initialize some data used later on
self.start_rotation = torch.tensor([1, 0, 0, 0], device=self._device, dtype=torch.float32)
self.up_vec = torch.tensor([0, 0, 1], dtype=torch.float32, device=self._device).repeat((self.num_envs, 1))
self.heading_vec = torch.tensor([1, 0, 0], dtype=torch.float32, device=self._device).repeat((self.num_envs, 1))
self.inv_start_rot = quat_conjugate(self.start_rotation).repeat((self.num_envs, 1))
self.basis_vec0 = self.heading_vec.clone()
self.basis_vec1 = self.up_vec.clone()
self.targets = torch.tensor([1000, 0, 0], dtype=torch.float32, device=self._device).repeat((self.num_envs, 1))
self.target_dirs = torch.tensor([1, 0, 0], dtype=torch.float32, device=self._device).repeat((self.num_envs, 1))
self.dt = 1.0 / 60.0
self.potentials = torch.tensor([-1000.0 / self.dt], dtype=torch.float32, device=self._device).repeat(
self.num_envs
)
self.prev_potentials = self.potentials.clone()
self.actions = torch.zeros((self.num_envs, self.num_actions), device=self._device)
# randomize all envs
indices = torch.arange(self._robots.count, dtype=torch.int64, device=self._device)
self.reset_idx(indices)
def calculate_metrics(self) -> None:
self.rew_buf[:] = calculate_metrics(
self.obs_buf,
self.actions,
self.up_weight,
self.heading_weight,
self.potentials,
self.prev_potentials,
self.actions_cost_scale,
self.energy_cost_scale,
self.termination_height,
self.death_cost,
self._robots.num_dof,
self.get_dof_at_limit_cost(),
self.alive_reward_scale,
self.motor_effort_ratio,
)
def is_done(self) -> None:
self.reset_buf[:] = is_done(
self.obs_buf, self.termination_height, self.reset_buf, self.progress_buf, self._max_episode_length
)
#####################################################################
###=========================jit functions=========================###
#####################################################################
@torch.jit.script
def normalize_angle(x):
return torch.atan2(torch.sin(x), torch.cos(x))
@torch.jit.script
def get_observations(
torso_position,
torso_rotation,
velocity,
ang_velocity,
dof_pos,
dof_vel,
targets,
potentials,
dt,
inv_start_rot,
basis_vec0,
basis_vec1,
dof_limits_lower,
dof_limits_upper,
dof_vel_scale,
sensor_force_torques,
num_envs,
contact_force_scale,
actions,
angular_velocity_scale,
):
# type: (Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, Tensor, float, Tensor, Tensor, Tensor, Tensor, Tensor, float, Tensor, int, float, Tensor, float) -> Tuple[Tensor, Tensor, Tensor, Tensor, Tensor]
to_target = targets - torso_position
to_target[:, 2] = 0.0
prev_potentials = potentials.clone()
potentials = -torch.norm(to_target, p=2, dim=-1) / dt
torso_quat, up_proj, heading_proj, up_vec, heading_vec = compute_heading_and_up(
torso_rotation, inv_start_rot, to_target, basis_vec0, basis_vec1, 2
)
vel_loc, angvel_loc, roll, pitch, yaw, angle_to_target = compute_rot(
torso_quat, velocity, ang_velocity, targets, torso_position
)
dof_pos_scaled = unscale(dof_pos, dof_limits_lower, dof_limits_upper)
# obs_buf shapes: 1, 3, 3, 1, 1, 1, 1, 1, num_dofs, num_dofs, num_sensors * 6, num_dofs
obs = torch.cat(
(
torso_position[:, 2].view(-1, 1),
vel_loc,
angvel_loc * angular_velocity_scale,
normalize_angle(yaw).unsqueeze(-1),
normalize_angle(roll).unsqueeze(-1),
normalize_angle(angle_to_target).unsqueeze(-1),
up_proj.unsqueeze(-1),
heading_proj.unsqueeze(-1),
dof_pos_scaled,
dof_vel * dof_vel_scale,
sensor_force_torques.reshape(num_envs, -1) * contact_force_scale,
actions,
),
dim=-1,
)
return obs, potentials, prev_potentials, up_vec, heading_vec
@torch.jit.script
def is_done(obs_buf, termination_height, reset_buf, progress_buf, max_episode_length):
# type: (Tensor, float, Tensor, Tensor, float) -> Tensor
reset = torch.where(obs_buf[:, 0] < termination_height, torch.ones_like(reset_buf), reset_buf)
reset = torch.where(progress_buf >= max_episode_length - 1, torch.ones_like(reset_buf), reset)
return reset
@torch.jit.script
def calculate_metrics(
obs_buf,
actions,
up_weight,
heading_weight,
potentials,
prev_potentials,
actions_cost_scale,
energy_cost_scale,
termination_height,
death_cost,
num_dof,
dof_at_limit_cost,
alive_reward_scale,
motor_effort_ratio,
):
# type: (Tensor, Tensor, float, float, Tensor, Tensor, float, float, float, float, int, Tensor, float, Tensor) -> Tensor
heading_weight_tensor = torch.ones_like(obs_buf[:, 11]) * heading_weight
heading_reward = torch.where(obs_buf[:, 11] > 0.8, heading_weight_tensor, heading_weight * obs_buf[:, 11] / 0.8)
# aligning up axis of robot and environment
up_reward = torch.zeros_like(heading_reward)
up_reward = torch.where(obs_buf[:, 10] > 0.93, up_reward + up_weight, up_reward)
# energy penalty for movement
actions_cost = torch.sum(actions**2, dim=-1)
electricity_cost = torch.sum(
torch.abs(actions * obs_buf[:, 12 + num_dof : 12 + num_dof * 2]) * motor_effort_ratio.unsqueeze(0), dim=-1
)
# reward for duration of staying alive
alive_reward = torch.ones_like(potentials) * alive_reward_scale
progress_reward = potentials - prev_potentials
total_reward = (
progress_reward
+ alive_reward
+ up_reward
+ heading_reward
- actions_cost_scale * actions_cost
- energy_cost_scale * electricity_cost
- dof_at_limit_cost
)
# adjust reward for fallen agents
total_reward = torch.where(
obs_buf[:, 0] < termination_height, torch.ones_like(total_reward) * death_cost, total_reward
)
return total_reward
| 13,249 | Python | 37.294798 | 214 | 0.628802 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/config.yaml |
# Task name - used to pick the class to load
task_name: ${task.name}
# experiment name. defaults to name of training config
experiment: ''
# if set to positive integer, overrides the default number of environments
num_envs: ''
# seed - set to -1 to choose random seed
seed: 42
# set to True for deterministic performance
torch_deterministic: False
# set the maximum number of learning iterations to train for. overrides default per-environment setting
max_iterations: ''
## Device config
physics_engine: 'physx'
# whether to use cpu or gpu pipeline
pipeline: 'gpu'
# whether to use cpu or gpu physx
sim_device: 'gpu'
# used for gpu simulation only - device id for running sim and task if pipeline=gpu
device_id: 0
# device to run RL
rl_device: 'cuda:0'
# multi-GPU training
multi_gpu: False
## PhysX arguments
num_threads: 4 # Number of worker threads per scene used by PhysX - for CPU PhysX only.
solver_type: 1 # 0: pgs, 1: tgs
# RLGames Arguments
# test - if set, run policy in inference mode (requires setting checkpoint to load)
test: False
# used to set checkpoint path
checkpoint: ''
# evaluate checkpoint
evaluation: False
# disables rendering
headless: False
# enables native livestream
enable_livestream: False
# timeout for MT script
mt_timeout: 90
wandb_activate: False
wandb_group: ''
wandb_name: ${train.params.config.name}
wandb_entity: ''
wandb_project: 'omniisaacgymenvs'
# path to a kit app file
kit_app: ''
# Warp
warp: False
# set default task and default training config based on task
defaults:
- _self_
- task: Cartpole
- train: ${task}PPO
- override hydra/job_logging: disabled
# set the directory where the output files get saved
hydra:
output_subdir: null
run:
dir: .
use_urdf: False
| 1,739 | YAML | 21.894737 | 103 | 0.738355 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/KukaKR120R2500ProReacher.yaml | # Ref: /omniisaacgymenvs/cfg/task/ShadowHand.yaml
# used to create the object
name: KukaKR120R2500ProReacher
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:2048,${...num_envs}}
envSpacing: 6
episodeLength: 600
clipObservations: 5.0
clipActions: 1.0
useRelativeControl: False
dofSpeedScale: 20.0
actionsMovingAverage: 1.0
controlFrequencyInv: 2 # 60 Hz
startPositionNoise: 0.01
startRotationNoise: 0.0
resetPositionNoise: 0.01
resetRotationNoise: 0.0
resetDofPosRandomInterval: 0.2
resetDofVelRandomInterval: 0.0
# Random forces applied to the object
forceScale: 0.0
forceProbRange: [0.001, 0.1]
forceDecay: 0.99
forceDecayInterval: 0.08
# reward -> dictionary
distRewardScale: -2.0
rotRewardScale: 1.0
rotEps: 0.1
actionPenaltyScale: -0.0002
reachGoalBonus: 250
velObsScale: 0.2
observationType: "full" # can only be "full"
successTolerance: 0.1
printNumSuccesses: False
maxConsecutiveSuccesses: 0
useURDF: ${resolve_default:True,${...use_urdf}}
sim:
dt: 0.0083 # 1/120 s
add_ground_plane: True
add_distant_light: False
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
# per-scene
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type} # 0: PGS, 1: TGS
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
# GPU buffers
gpu_max_rigid_contact_count: 1048576
gpu_max_rigid_patch_count: 33554432
gpu_found_lost_pairs_capacity: 20971520
gpu_found_lost_aggregate_pairs_capacity: 20971520
gpu_total_aggregate_pairs_capacity: 20971520
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 33554432
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
kuka:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: False
object:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.000
stabilization_threshold: 0.0025
# per-body
density: -1
max_depenetration_velocity: 1000.0
goal_object:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.000
stabilization_threshold: 0.0025
# per-body
density: -1
max_depenetration_velocity: 1000.0
sim2real:
enabled: False
fail_quietely: False
verbose: False
| 3,304 | YAML | 25.023622 | 71 | 0.70339 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/CartpoleCamera.yaml | defaults:
- Cartpole
- _self_
name: CartpoleCamera
env:
numEnvs: ${resolve_default:32,${...num_envs}}
envSpacing: 20.0
cameraWidth: 240
cameraHeight: 160
exportImages: False
sim:
rendering_dt: 0.0166 # 1/60
# set to True if you use camera sensors in the environment
enable_cameras: True
add_ground_plane: False
add_distant_light: True
| 363 | YAML | 16.333333 | 60 | 0.69697 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/FrankaDeformable.yaml | # used to create the object
name: FrankaDeformable
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:1024,${...num_envs}} # 2048#4096
envSpacing: 3.0
episodeLength: 100 # 150 #350 #500
enableDebugVis: False
clipObservations: 5.0
clipActions: 1.0
controlFrequencyInv: 2 # 60 Hz
startPositionNoise: 0.0
startRotationNoise: 0.0
numProps: 4
aggregateMode: 3
actionScale: 7.5
dofVelocityScale: 0.1
distRewardScale: 2.0
rotRewardScale: 0.5
aroundHandleRewardScale: 10.0
openRewardScale: 7.5
fingerDistRewardScale: 100.0
actionPenaltyScale: 0.01
fingerCloseRewardScale: 10.0
sim:
dt: 0.0083 # 1/120 s
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 8 # 12
solver_velocity_iteration_count: 0 # 1
contact_offset: 0.02 #0.005
rest_offset: 0.001
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 1000.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 33554432
gpu_found_lost_pairs_capacity: 524288 #20965884
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 1048576
gpu_max_soft_body_contacts: 4194304 #2097152 #16777216 #8388608 #2097152 #1048576
gpu_max_particle_contacts: 1048576 #2097152 #1048576
gpu_heap_capacity: 33554432
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
franka:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
beaker:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
cube:
# -1 to use default values
override_usd_defaults: False
make_kinematic: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
# # per-shape
# contact_offset: 0.02
# rest_offset: 0.001
| 3,421 | YAML | 25.944882 | 85 | 0.691903 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/FrankaCabinet.yaml | # used to create the object
name: FrankaCabinet
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 3.0
episodeLength: 500
enableDebugVis: False
clipObservations: 5.0
clipActions: 1.0
controlFrequencyInv: 2 # 60 Hz
startPositionNoise: 0.0
startRotationNoise: 0.0
numProps: 4
aggregateMode: 3
actionScale: 7.5
dofVelocityScale: 0.1
distRewardScale: 2.0
rotRewardScale: 0.5
aroundHandleRewardScale: 10.0
openRewardScale: 7.5
fingerDistRewardScale: 100.0
actionPenaltyScale: 0.01
fingerCloseRewardScale: 10.0
sim:
dt: 0.0083 # 1/120 s
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
contact_offset: 0.005
rest_offset: 0.0
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 1000.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 33554432
gpu_found_lost_pairs_capacity: 524288
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 1048576
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 33554432
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
franka:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
cabinet:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
sleep_threshold: 0.0
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
prop:
# -1 to use default values
override_usd_defaults: False
make_kinematic: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 12
solver_velocity_iteration_count: 1
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: 100
max_depenetration_velocity: 1000.0
# per-shape
contact_offset: 0.005
rest_offset: 0.0
| 3,287 | YAML | 25.304 | 71 | 0.695467 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Ant.yaml | # used to create the object
name: Ant
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
# numEnvs: ${...num_envs}
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 5
episodeLength: 1000
enableDebugVis: False
clipActions: 1.0
powerScale: 0.5
controlFrequencyInv: 2 # 60 Hz
# reward parameters
headingWeight: 0.5
upWeight: 0.1
# cost parameters
actionsCost: 0.005
energyCost: 0.05
dofVelocityScale: 0.2
angularVelocityScale: 1.0
contactForceScale: 0.1
jointsAtLimitCost: 0.1
deathCost: -2.0
terminationHeight: 0.31
alive_reward_scale: 0.5
sim:
dt: 0.0083 # 1/120 s
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.0
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 10.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 8192
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 8192
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
Ant:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 10.0 | 2,370 | YAML | 24.771739 | 71 | 0.690717 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/AnymalTerrain.yaml | name: AnymalTerrain
physics_engine: ${..physics_engine}
env:
numEnvs: ${resolve_default:2048,${...num_envs}}
numObservations: 188
numActions: 12
envSpacing: 3. # [m]
terrain:
staticFriction: 1.0 # [-]
dynamicFriction: 1.0 # [-]
restitution: 0. # [-]
# rough terrain only:
curriculum: true
maxInitMapLevel: 0
mapLength: 8.
mapWidth: 8.
numLevels: 10
numTerrains: 20
# terrain types: [smooth slope, rough slope, stairs up, stairs down, discrete]
terrainProportions: [0.1, 0.1, 0.35, 0.25, 0.2]
# tri mesh only:
slopeTreshold: 0.5
baseInitState:
pos: [0.0, 0.0, 0.62] # x,y,z [m]
rot: [1.0, 0.0, 0.0, 0.0] # w,x,y,z [quat]
vLinear: [0.0, 0.0, 0.0] # x,y,z [m/s]
vAngular: [0.0, 0.0, 0.0] # x,y,z [rad/s]
randomCommandVelocityRanges:
# train
linear_x: [-1., 1.] # min max [m/s]
linear_y: [-1., 1.] # min max [m/s]
yaw: [-3.14, 3.14] # min max [rad/s]
control:
# PD Drive parameters:
stiffness: 80.0 # [N*m/rad]
damping: 2.0 # [N*m*s/rad]
# action scale: target angle = actionScale * action + defaultAngle
actionScale: 0.5
# decimation: Number of control action updates @ sim DT per policy DT
decimation: 4
defaultJointAngles: # = target angles when action = 0.0
LF_HAA: 0.03 # [rad]
LH_HAA: 0.03 # [rad]
RF_HAA: -0.03 # [rad]
RH_HAA: -0.03 # [rad]
LF_HFE: 0.4 # [rad]
LH_HFE: -0.4 # [rad]
RF_HFE: 0.4 # [rad]
RH_HFE: -0.4 # [rad]
LF_KFE: -0.8 # [rad]
LH_KFE: 0.8 # [rad]
RF_KFE: -0.8 # [rad]
RH_KFE: 0.8 # [rad]
learn:
# rewards
terminalReward: 0.0
linearVelocityXYRewardScale: 1.0
linearVelocityZRewardScale: -4.0
angularVelocityXYRewardScale: -0.05
angularVelocityZRewardScale: 0.5
orientationRewardScale: -0.
torqueRewardScale: -0.00002
jointAccRewardScale: -0.0005
baseHeightRewardScale: -0.0
actionRateRewardScale: -0.01
fallenOverRewardScale: -1.0
# cosmetics
hipRewardScale: -0. #25
# normalization
linearVelocityScale: 2.0
angularVelocityScale: 0.25
dofPositionScale: 1.0
dofVelocityScale: 0.05
heightMeasurementScale: 5.0
# noise
addNoise: true
noiseLevel: 1.0 # scales other values
dofPositionNoise: 0.01
dofVelocityNoise: 1.5
linearVelocityNoise: 0.1
angularVelocityNoise: 0.2
gravityNoise: 0.05
heightMeasurementNoise: 0.06
#randomization
pushInterval_s: 15
# episode length in seconds
episodeLength_s: 20
sim:
dt: 0.005
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: False
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: True
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.0
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 100.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 163840
gpu_found_lost_pairs_capacity: 4194304
gpu_found_lost_aggregate_pairs_capacity: 33554432
gpu_total_aggregate_pairs_capacity: 4194304
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 134217728
gpu_temp_buffer_capacity: 33554432
gpu_max_num_partitions: 8
anymal:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: False
# also in stage params
# per-actor
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 100.0
| 4,346 | YAML | 25.345454 | 82 | 0.633916 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/BallBalance.yaml | # used to create the object
name: BallBalance
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 2.0
maxEpisodeLength: 600
actionSpeedScale: 20
clipObservations: 5.0
clipActions: 1.0
sim:
dt: 0.01
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.001
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 1000.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 262144
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 262144
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
table:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
ball:
# -1 to use default values
override_usd_defaults: False
make_kinematic: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: 200
max_depenetration_velocity: 1000.0
| 2,458 | YAML | 25.728261 | 71 | 0.690806 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/FactoryBase.yaml | # See schema in factory_schema_config_base.py for descriptions of parameters.
defaults:
- _self_
- /factory_schema_config_base
sim:
add_damping: True
disable_contact_processing: False
env:
env_spacing: 1.5
franka_depth: 0.5
table_height: 0.4
franka_friction: 1.0
table_friction: 0.3
| 309 | YAML | 16.222221 | 77 | 0.699029 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Humanoid.yaml | # used to create the object
name: Humanoid
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
# numEnvs: ${...num_envs}
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 5
episodeLength: 1000
enableDebugVis: False
clipActions: 1.0
powerScale: 1.0
controlFrequencyInv: 2 # 60 Hz
# reward parameters
headingWeight: 0.5
upWeight: 0.1
# cost parameters
actionsCost: 0.01
energyCost: 0.05
dofVelocityScale: 0.1
angularVelocityScale: 0.25
contactForceScale: 0.01
jointsAtLimitCost: 0.25
deathCost: -1.0
terminationHeight: 0.8
alive_reward_scale: 2.0
sim:
dt: 0.0083 # 1/120 s
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 10.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 8192
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 8192
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
Humanoid:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 10.0
| 2,335 | YAML | 24.670329 | 71 | 0.693362 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/AllegroHand.yaml | # used to create the object
name: AllegroHand
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:8192,${...num_envs}}
envSpacing: 0.75
episodeLength: 600
clipObservations: 5.0
clipActions: 1.0
useRelativeControl: False
dofSpeedScale: 20.0
actionsMovingAverage: 1.0
controlFrequencyInv: 4 # 30 Hz
startPositionNoise: 0.01
startRotationNoise: 0.0
resetPositionNoise: 0.01
resetRotationNoise: 0.0
resetDofPosRandomInterval: 0.2
resetDofVelRandomInterval: 0.0
# reward -> dictionary
distRewardScale: -10.0
rotRewardScale: 1.0
rotEps: 0.1
actionPenaltyScale: -0.0002
reachGoalBonus: 250
fallDistance: 0.24
fallPenalty: 0.0
velObsScale: 0.2
objectType: "block"
observationType: "full" # can be "full_no_vel", "full"
successTolerance: 0.1
printNumSuccesses: False
maxConsecutiveSuccesses: 0
sim:
dt: 0.0083 # 1/120 s
add_ground_plane: True
add_distant_light: False
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
# per-scene
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type} # 0: PGS, 1: TGS
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 33554432
gpu_found_lost_pairs_capacity: 819200
gpu_found_lost_aggregate_pairs_capacity: 819200
gpu_total_aggregate_pairs_capacity: 1048576
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 33554432
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
allegro_hand:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: False
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.0005
# per-body
density: -1
max_depenetration_velocity: 1000.0
object:
# -1 to use default values
override_usd_defaults: False
make_kinematic: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.0025
# per-body
density: 400.0
max_depenetration_velocity: 1000.0
goal_object:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 8
solver_velocity_iteration_count: 0
sleep_threshold: 0.000
stabilization_threshold: 0.0025
# per-body
density: -1
max_depenetration_velocity: 1000.0
| 3,360 | YAML | 25.464567 | 71 | 0.69881 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/HumanoidSAC.yaml | # used to create the object
defaults:
- Humanoid
- _self_
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:64,${...num_envs}} | 168 | YAML | 20.124997 | 52 | 0.678571 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Ingenuity.yaml | # used to create the object
name: Ingenuity
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 2.5
maxEpisodeLength: 2000
enableDebugVis: False
clipObservations: 5.0
clipActions: 1.0
sim:
dt: 0.01
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -3.721]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
# set to True if you use camera sensors in the environment
enable_cameras: False
disable_contact_processing: False
physx:
num_threads: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.001
bounce_threshold_velocity: 0.2
max_depenetration_velocity: 1000.0
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: False
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 4194304
gpu_found_lost_aggregate_pairs_capacity: 33554432
gpu_total_aggregate_pairs_capacity: 4194304
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
ingenuity:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
ball:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0 | 2,351 | YAML | 27 | 71 | 0.693322 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Quadcopter.yaml | # used to create the object
name: Quadcopter
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 1.25
maxEpisodeLength: 500
enableDebugVis: False
clipObservations: 5.0
clipActions: 1.0
sim:
dt: 0.01
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.001
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 1000.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 8192
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 8192
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
copter:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
ball:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
| 2,452 | YAML | 25.663043 | 71 | 0.690457 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Crazyflie.yaml | # used to create the object
name: Crazyflie
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 2.5
maxEpisodeLength: 700
enableDebugVis: False
clipObservations: 5.0
clipActions: 1.0
sim:
dt: 0.01
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
# set to True if you use camera sensors in the environment
enable_cameras: False
disable_contact_processing: False
physx:
num_threads: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.001
bounce_threshold_velocity: 0.2
max_depenetration_velocity: 1000.0
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: False
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 4194304
gpu_found_lost_aggregate_pairs_capacity: 33554432
gpu_total_aggregate_pairs_capacity: 4194304
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
crazyflie:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: True
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
ball:
# -1 to use default values
override_usd_defaults: False
make_kinematic: True
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 6
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 1000.0
| 2,350 | YAML | 26.658823 | 71 | 0.692766 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/FactoryEnvNutBolt.yaml | # See schema in factory_schema_config_env.py for descriptions of common parameters.
defaults:
- _self_
- /factory_schema_config_env
sim:
disable_franka_collisions: False
disable_nut_collisions: False
disable_bolt_collisions: False
disable_contact_processing: False
env:
env_name: 'FactoryEnvNutBolt'
desired_subassemblies: ['nut_bolt_m16', 'nut_bolt_m16']
nut_lateral_offset: 0.1 # Y-axis offset of nut before initial reset to prevent initial interpenetration with bolt
nut_bolt_density: 7850.0
nut_bolt_friction: 0.3
# Subassembly options:
# {nut_bolt_m4, nut_bolt_m8, nut_bolt_m12, nut_bolt_m16, nut_bolt_m20}
| 643 | YAML | 28.272726 | 116 | 0.73717 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/AntSAC.yaml | # used to create the object
defaults:
- Ant
- _self_
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:64,${...num_envs}} | 163 | YAML | 19.499998 | 52 | 0.668712 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Cartpole.yaml | # used to create the object
name: Cartpole
physics_engine: ${..physics_engine}
# if given, will override the device setting in gym.
env:
numEnvs: ${resolve_default:512,${...num_envs}}
envSpacing: 4.0
resetDist: 3.0
maxEffort: 400.0
clipObservations: 5.0
clipActions: 1.0
controlFrequencyInv: 2 # 60 Hz
sim:
dt: 0.0083 # 1/120 s
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
contact_offset: 0.02
rest_offset: 0.001
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 100.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 81920
gpu_found_lost_pairs_capacity: 1024
gpu_found_lost_aggregate_pairs_capacity: 262144
gpu_total_aggregate_pairs_capacity: 1024
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 67108864
gpu_temp_buffer_capacity: 16777216
gpu_max_num_partitions: 8
Cartpole:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 4
solver_velocity_iteration_count: 0
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 100.0
# per-shape
contact_offset: 0.02
rest_offset: 0.001 | 2,124 | YAML | 26.243589 | 71 | 0.686911 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/Anymal.yaml | # used to create the object
name: Anymal
physics_engine: ${..physics_engine}
env:
numEnvs: ${resolve_default:4096,${...num_envs}}
envSpacing: 4. # [m]
clipObservations: 5.0
clipActions: 1.0
controlFrequencyInv: 2
baseInitState:
pos: [0.0, 0.0, 0.62] # x,y,z [m]
rot: [0.0, 0.0, 0.0, 1.0] # x,y,z,w [quat]
vLinear: [0.0, 0.0, 0.0] # x,y,z [m/s]
vAngular: [0.0, 0.0, 0.0] # x,y,z [rad/s]
randomCommandVelocityRanges:
linear_x: [-2., 2.] # min max [m/s]
linear_y: [-1., 1.] # min max [m/s]
yaw: [-1., 1.] # min max [rad/s]
control:
# PD Drive parameters:
stiffness: 85.0 # [N*m/rad]
damping: 2.0 # [N*m*s/rad]
actionScale: 13.5
defaultJointAngles: # = target angles when action = 0.0
LF_HAA: 0.03 # [rad]
LH_HAA: 0.03 # [rad]
RF_HAA: -0.03 # [rad]
RH_HAA: -0.03 # [rad]
LF_HFE: 0.4 # [rad]
LH_HFE: -0.4 # [rad]
RF_HFE: 0.4 # [rad]
RH_HFE: -0.4 # [rad]
LF_KFE: -0.8 # [rad]
LH_KFE: 0.8 # [rad]
RF_KFE: -0.8 # [rad]
RH_KFE: 0.8 # [rad]
learn:
# rewards
linearVelocityXYRewardScale: 1.0
angularVelocityZRewardScale: 0.5
linearVelocityZRewardScale: -0.03
jointAccRewardScale: -0.0003
actionRateRewardScale: -0.006
cosmeticRewardScale: -0.06
# normalization
linearVelocityScale: 2.0
angularVelocityScale: 0.25
dofPositionScale: 1.0
dofVelocityScale: 0.05
# episode length in seconds
episodeLength_s: 50
sim:
dt: 0.01
use_gpu_pipeline: ${eq:${...pipeline},"gpu"}
gravity: [0.0, 0.0, -9.81]
add_ground_plane: True
add_distant_light: False
use_fabric: True
enable_scene_query_support: False
disable_contact_processing: False
# set to True if you use camera sensors in the environment
enable_cameras: False
default_physics_material:
static_friction: 1.0
dynamic_friction: 1.0
restitution: 0.0
physx:
worker_thread_count: ${....num_threads}
solver_type: ${....solver_type}
use_gpu: ${eq:${....sim_device},"gpu"} # set to False to run on CPU
solver_position_iteration_count: 4
solver_velocity_iteration_count: 1
contact_offset: 0.02
rest_offset: 0.0
bounce_threshold_velocity: 0.2
friction_offset_threshold: 0.04
friction_correlation_distance: 0.025
enable_sleeping: True
enable_stabilization: True
max_depenetration_velocity: 100.0
# GPU buffers
gpu_max_rigid_contact_count: 524288
gpu_max_rigid_patch_count: 163840
gpu_found_lost_pairs_capacity: 4194304
gpu_found_lost_aggregate_pairs_capacity: 33554432
gpu_total_aggregate_pairs_capacity: 4194304
gpu_max_soft_body_contacts: 1048576
gpu_max_particle_contacts: 1048576
gpu_heap_capacity: 134217728
gpu_temp_buffer_capacity: 33554432
gpu_max_num_partitions: 8
Anymal:
# -1 to use default values
override_usd_defaults: False
enable_self_collisions: False
enable_gyroscopic_forces: True
# also in stage params
# per-actor
solver_position_iteration_count: 4
solver_velocity_iteration_count: 1
sleep_threshold: 0.005
stabilization_threshold: 0.001
# per-body
density: -1
max_depenetration_velocity: 100.0
| 3,270 | YAML | 24.960317 | 71 | 0.626911 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/task/ShadowHandOpenAI_LSTM.yaml | # specifies what the config is when running `ShadowHandOpenAI` in LSTM mode
defaults:
- ShadowHandOpenAI_FF
- _self_
env:
numEnvs: ${resolve_default:8192,${...num_envs}}
| 178 | YAML | 18.888887 | 75 | 0.707865 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/train/ShadowHandOpenAI_FFPPO.yaml | params:
seed: ${...seed}
algo:
name: a2c_continuous
model:
name: continuous_a2c_logstd
network:
name: actor_critic
separate: False
space:
continuous:
mu_activation: None
sigma_activation: None
mu_init:
name: default
sigma_init:
name: const_initializer
val: 0
fixed_sigma: True
mlp:
units: [400, 400, 200, 100]
activation: elu
d2rl: False
initializer:
name: default
regularizer:
name: None
load_checkpoint: ${if:${...checkpoint},True,False}
load_path: ${...checkpoint}
config:
name: ${resolve_default:ShadowHandOpenAI_FF,${....experiment}}
full_experiment_name: ${.name}
device: ${....rl_device}
device_name: ${....rl_device}
env_name: rlgpu
multi_gpu: ${....multi_gpu}
ppo: True
mixed_precision: False
normalize_input: True
normalize_value: True
num_actors: ${....task.env.numEnvs}
reward_shaper:
scale_value: 0.01
normalize_advantage: True
gamma: 0.998
tau: 0.95
learning_rate: 5e-4
lr_schedule: adaptive
schedule_type: standard
kl_threshold: 0.016
score_to_win: 100000
max_epochs: ${resolve_default:10000,${....max_iterations}}
save_best_after: 100
save_frequency: 200
print_stats: True
grad_norm: 1.0
entropy_coef: 0.0
truncate_grads: True
e_clip: 0.2
horizon_length: 16
minibatch_size: 16384
mini_epochs: 4
critic_coef: 4
clip_value: True
seq_length: 4
bounds_loss_coef: 0.0001
central_value_config:
minibatch_size: 32864
mini_epochs: 4
learning_rate: 5e-4
lr_schedule: adaptive
schedule_type: standard
kl_threshold: 0.016
clip_value: True
normalize_input: True
truncate_grads: True
network:
name: actor_critic
central_value: True
mlp:
units: [512, 512, 256, 128]
activation: elu
d2rl: False
initializer:
name: default
regularizer:
name: None
player:
deterministic: True
games_num: 100000
print_stats: True
| 2,215 | YAML | 20.940594 | 66 | 0.577427 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/train/AnymalTerrainPPO.yaml | params:
seed: ${...seed}
algo:
name: a2c_continuous
model:
name: continuous_a2c_logstd
network:
name: actor_critic
separate: True
space:
continuous:
mu_activation: None
sigma_activation: None
mu_init:
name: default
sigma_init:
name: const_initializer
val: 0. # std = 1.
fixed_sigma: True
mlp:
units: [512, 256, 128]
activation: elu
d2rl: False
initializer:
name: default
regularizer:
name: None
# rnn:
# name: lstm
# units: 128
# layers: 1
# before_mlp: True
# concat_input: True
# layer_norm: False
load_checkpoint: ${if:${...checkpoint},True,False} # flag which sets whether to load the checkpoint
load_path: ${...checkpoint} # path to the checkpoint to load
config:
name: ${resolve_default:AnymalTerrain,${....experiment}}
full_experiment_name: ${.name}
device: ${....rl_device}
device_name: ${....rl_device}
env_name: rlgpu
multi_gpu: ${....multi_gpu}
ppo: True
mixed_precision: False # True
normalize_input: True
normalize_value: True
normalize_advantage: True
value_bootstrap: True
clip_actions: False
num_actors: ${....task.env.numEnvs}
reward_shaper:
scale_value: 1.0
gamma: 0.99
tau: 0.95
e_clip: 0.2
entropy_coef: 0.001
learning_rate: 3.e-4 # overwritten by adaptive lr_schedule
lr_schedule: adaptive
kl_threshold: 0.008 # target kl for adaptive lr
truncate_grads: True
grad_norm: 1.
horizon_length: 48
minibatch_size: 16384
mini_epochs: 5
critic_coef: 2
clip_value: True
seq_length: 4 # only for rnn
bounds_loss_coef: 0.
max_epochs: ${resolve_default:2000,${....max_iterations}}
save_best_after: 100
score_to_win: 20000
save_frequency: 50
print_stats: True
| 1,928 | YAML | 21.694117 | 101 | 0.592842 |
j3soon/OmniIsaacGymEnvs-KukaReacher/omniisaacgymenvs/cfg/train/HumanoidPPO.yaml | params:
seed: ${...seed}
algo:
name: a2c_continuous
model:
name: continuous_a2c_logstd
network:
name: actor_critic
separate: False
space:
continuous:
mu_activation: None
sigma_activation: None
mu_init:
name: default
sigma_init:
name: const_initializer
val: 0
fixed_sigma: True
mlp:
units: [400, 200, 100]
activation: elu
d2rl: False
initializer:
name: default
regularizer:
name: None
load_checkpoint: ${if:${...checkpoint},True,False} # flag which sets whether to load the checkpoint
load_path: ${...checkpoint} # path to the checkpoint to load
config:
name: ${resolve_default:Humanoid,${....experiment}}
full_experiment_name: ${.name}
env_name: rlgpu
device: ${....rl_device}
device_name: ${....rl_device}
multi_gpu: ${....multi_gpu}
ppo: True
mixed_precision: True
normalize_input: True
normalize_value: True
value_bootstrap: True
num_actors: ${....task.env.numEnvs}
reward_shaper:
scale_value: 0.01
normalize_advantage: True
gamma: 0.99
tau: 0.95
learning_rate: 5e-4
lr_schedule: adaptive
kl_threshold: 0.008
score_to_win: 20000
max_epochs: ${resolve_default:1000,${....max_iterations}}
save_best_after: 100
save_frequency: 100
grad_norm: 1.0
entropy_coef: 0.0
truncate_grads: True
e_clip: 0.2
horizon_length: 32
minibatch_size: 32768
mini_epochs: 5
critic_coef: 4
clip_value: True
seq_length: 4
bounds_loss_coef: 0.0001
| 1,639 | YAML | 21.465753 | 101 | 0.594875 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.