import numpy as np import os import IPython from cliport import tasks from cliport.dataset import RavensDataset from cliport.environments.environment import Environment from pygments import highlight from pygments.lexers import PythonLexer from pygments.formatters import TerminalFormatter import time import random import json import traceback from gensim.utils import ( mkdir_if_missing, save_text, save_stat, compute_diversity_score_from_assets, add_to_txt ) import pybullet as p class SimulationRunner: """ the main class that runs simulation loop """ def __init__(self, cfg, agent, critic, memory): self.cfg = cfg self.agent = agent self.critic = critic self.memory = memory # statistics self.syntax_pass_rate = 0 self.runtime_pass_rate = 0 self.env_pass_rate = 0 self.curr_trials = 0 self.prompt_folder = f"prompts/{cfg['prompt_folder']}" self.chat_log = memory.chat_log self.task_asset_logs = [] # All the generated tasks in this run. # Different from the ones in online buffer that can load from offline. self.generated_task_assets = [] self.generated_task_programs = [] self.generated_task_names = [] self.generated_tasks = [] self.passed_tasks = [] # accepted ones def print_current_stats(self): """ print the current statistics of the simulation design """ print("=========================================================") print(f"{self.cfg['prompt_folder']} Trial {self.curr_trials} SYNTAX_PASS_RATE: {(self.syntax_pass_rate / (self.curr_trials)) * 100:.1f}% RUNTIME_PASS_RATE: {(self.runtime_pass_rate / (self.curr_trials)) * 100:.1f}% ENV_PASS_RATE: {(self.env_pass_rate / (self.curr_trials)) * 100:.1f}%") print("=========================================================") def save_stats(self): """ save the final simulation statistics """ self.diversity_score = compute_diversity_score_from_assets(self.task_asset_logs, self.curr_trials) save_stat(self.cfg, self.cfg['model_output_dir'], self.generated_tasks, self.syntax_pass_rate / (self.curr_trials), self.runtime_pass_rate / (self.curr_trials), self.env_pass_rate / (self.curr_trials), self.diversity_score) print("Model Folder: ", self.cfg['model_output_dir']) print(f"Total {len(self.generated_tasks)} New Tasks:", [task['task-name'] for task in self.generated_tasks]) try: print(f"Added {len(self.passed_tasks)} Tasks:", self.passed_tasks) except: pass def example_task_creation(self): """ create the task through interactions of agent and critic """ self.task_creation_pass = True mkdir_if_missing(self.cfg['model_output_dir']) try: start_time = time.time() self.generated_task = {'task-name': 'TASK_NAME_TEMPLATE', 'task-description': 'TASK_STRING_TEMPLATE', 'assets-used': ['ASSET_1', 'ASSET_2', Ellipsis]} print("generated_task\n", self.generated_task) yield "Task Generated ==>", None, None self.generated_asset = self.agent.propose_assets() # self.generated_asset = {} print("generated_asset\n", self.generated_asset) yield "Task Generated ==> Asset Generated ==> ", None, None yield "Task Generated ==> Asset Generated ==> API Reviewed ==> ", None, None yield "Task Generated ==> Asset Generated ==> API Reviewed ==> Error Reviewed ==> ", None, None self.curr_task_name = self.generated_task_name = 'BuildWheel' self.generated_code = """ import numpy as np from cliport.tasks.task import Task from cliport.utils import utils class BuildWheel(Task): def __init__(self): super().__init__() self.max_steps = 10 self.lang_template = "Construct a wheel using blocks and a sphere. First, position eight blocks in a circular layout on the tabletop. Each block should be touching its two neighbors and colored in alternating red and blue. Then place a green sphere in the center of the circular layout, completing the wheel." self.task_completed_desc = "done building wheel." self.additional_reset() def reset(self, env): super().reset(env) # Add blocks. block_size = (0.04, 0.04, 0.04) block_urdf = 'block/block.urdf' block_colors = [utils.COLORS['red'], utils.COLORS['blue']] blocks = [] for i in range(8): block_pose = self.get_random_pose(env, block_size) block_id = env.add_object(block_urdf, block_pose, color=block_colors[i % 2]) blocks.append(block_id) # Add sphere. sphere_size = (0.04, 0.04, 0.04) sphere_urdf = 'sphere/sphere.urdf' sphere_color = utils.COLORS['green'] sphere_pose = ((0.5, 0.0, 0.0), (0,0,0,1)) # fixed pose sphere_id = env.add_object(sphere_urdf, sphere_pose, color=sphere_color) # Goal: blocks are arranged in a circle and sphere is in the center. circle_radius = 0.1 circle_center = (0, 0, block_size[2] / 2) angles = np.linspace(0, 2 * np.pi, 8, endpoint=False) block_poses = [(circle_center[0] + circle_radius * np.cos(angle), circle_center[1] + circle_radius * np.sin(angle), circle_center[2]) for angle in angles] block_poses = [(utils.apply(sphere_pose, pos), sphere_pose[1]) for pos in block_poses] self.add_goal(objs=blocks, matches=np.ones((8, 8)), targ_poses=block_poses, replace=False, rotations=True, metric='pose', params=None, step_max_reward=8 / 9) # Goal: sphere is in the center of the blocks. self.add_goal(objs=[sphere_id], matches=np.ones((1, 1)), targ_poses=[sphere_pose], replace=False, rotations=False, metric='pose', params=None, step_max_reward=1 / 9) self.lang_goals.append(self.lang_template) """ print("generated_code\n", self.generated_code) print("curr_task_name\n", self.curr_task_name) yield "Task Generated ==> Asset Generated ==> API Reviewed ==> Error Reviewed ==> Code Generated ==> ", self.generated_code, None self.generated_tasks.append(self.generated_task) self.generated_task_assets.append(self.generated_asset) self.generated_task_programs.append(self.generated_code) self.generated_task_names.append(self.generated_task_name) except: to_print = highlight(f"{str(traceback.format_exc())}", PythonLexer(), TerminalFormatter()) print("Task Creation Exception:", to_print) self.task_creation_pass = False # self.curr_task_name = self.generated_task['task-name'] print("task creation time {:.3f}".format(time.time() - start_time)) def task_creation(self): """ create the task through interactions of agent and critic """ self.task_creation_pass = True mkdir_if_missing(self.cfg['model_output_dir']) try: start_time = time.time() self.generated_task = self.agent.propose_task(self.generated_task_names) # self.generated_task = {'task-name': 'TASK_NAME_TEMPLATE', 'task-description': 'TASK_STRING_TEMPLATE', 'assets-used': ['ASSET_1', 'ASSET_2', Ellipsis]} print("generated_task\n", self.generated_task) yield "Task Generated ==>", None, None self.generated_asset = self.agent.propose_assets() # self.generated_asset = {} print("generated_asset\n", self.generated_asset) yield "Task Generated ==> Asset Generated ==> ", None, None self.agent.api_review() yield "Task Generated ==> Asset Generated ==> API Reviewed ==> ", None, None self.critic.error_review(self.generated_task) yield "Task Generated ==> Asset Generated ==> API Reviewed ==> Error Reviewed ==> ", None, None self.generated_code, self.curr_task_name = self.agent.implement_task() self.task_asset_logs.append(self.generated_task["assets-used"]) self.generated_task_name = self.generated_task["task-name"] # self.curr_task_name = self.generated_task_name = 'BuildWheel' # # self.generated_code = """ # import numpy as np # from cliport.tasks.task import Task # from cliport.utils import utils # # class BuildWheel(Task): # # def __init__(self): # super().__init__() # self.max_steps = 10 # self.lang_template = "Construct a wheel using blocks and a sphere. First, position eight blocks in a circular layout on the tabletop. Each block should be touching its two neighbors and colored in alternating red and blue. Then place a green sphere in the center of the circular layout, completing the wheel." # self.task_completed_desc = "done building wheel." # self.additional_reset() # # def reset(self, env): # super().reset(env) # # # Add blocks. # block_size = (0.04, 0.04, 0.04) # block_urdf = 'block/block.urdf' # block_colors = [utils.COLORS['red'], utils.COLORS['blue']] # blocks = [] # for i in range(8): # block_pose = self.get_random_pose(env, block_size) # block_id = env.add_object(block_urdf, block_pose, color=block_colors[i % 2]) # blocks.append(block_id) # # # Add sphere. # sphere_size = (0.04, 0.04, 0.04) # sphere_urdf = 'sphere/sphere.urdf' # sphere_color = utils.COLORS['green'] # sphere_pose = ((0.5, 0.0, 0.0), (0,0,0,1)) # fixed pose # sphere_id = env.add_object(sphere_urdf, sphere_pose, color=sphere_color) # # # Goal: blocks are arranged in a circle and sphere is in the center. # circle_radius = 0.1 # circle_center = (0, 0, block_size[2] / 2) # angles = np.linspace(0, 2 * np.pi, 8, endpoint=False) # block_poses = [(circle_center[0] + circle_radius * np.cos(angle), # circle_center[1] + circle_radius * np.sin(angle), # circle_center[2]) for angle in angles] # block_poses = [(utils.apply(sphere_pose, pos), sphere_pose[1]) for pos in block_poses] # self.add_goal(objs=blocks, matches=np.ones((8, 8)), targ_poses=block_poses, replace=False, # rotations=True, metric='pose', params=None, step_max_reward=8 / 9) # # # Goal: sphere is in the center of the blocks. # self.add_goal(objs=[sphere_id], matches=np.ones((1, 1)), targ_poses=[sphere_pose], replace=False, # rotations=False, metric='pose', params=None, step_max_reward=1 / 9) # # self.lang_goals.append(self.lang_template) # """ print("generated_code\n", self.generated_code) print("curr_task_name\n", self.curr_task_name) yield "Task Generated ==> Asset Generated ==> API Reviewed ==> Error Reviewed ==> Code Generated ==> ", self.generated_code, None self.generated_tasks.append(self.generated_task) self.generated_task_assets.append(self.generated_asset) self.generated_task_programs.append(self.generated_code) self.generated_task_names.append(self.generated_task_name) except: to_print = highlight(f"{str(traceback.format_exc())}", PythonLexer(), TerminalFormatter()) print("Task Creation Exception:", to_print) self.task_creation_pass = False # self.curr_task_name = self.generated_task['task-name'] print("task creation time {:.3f}".format(time.time() - start_time)) def setup_env(self): """ build the new task""" env = Environment( self.cfg['assets_root'], disp=self.cfg['disp'], shared_memory=self.cfg['shared_memory'], hz=480, record_cfg=self.cfg['record'] ) task = eval(self.curr_task_name)() task.mode = self.cfg['mode'] record = self.cfg['record']['save_video'] save_data = self.cfg['save_data'] # Initialize scripted oracle agent and dataset. expert = task.oracle(env) self.cfg['task'] = self.generated_task["task-name"] data_path = os.path.join(self.cfg['data_dir'], "{}-{}".format(self.generated_task["task-name"], task.mode)) dataset = RavensDataset(data_path, self.cfg, n_demos=0, augment=False) print(f"Saving to: {data_path}") print(f"Mode: {task.mode}") # Start video recording if record: env.start_rec(f'{dataset.n_episodes+1:06d}') return task, dataset, env, expert def run_one_episode(self, dataset, expert, env, task, episode, seed): """ run the new task for one episode """ add_to_txt( self.chat_log, f"================= TRIAL: {self.curr_trials}", with_print=True) record = self.cfg['record']['save_video'] np.random.seed(seed) random.seed(seed) print('Oracle demo: {}/{} | Seed: {}'.format(dataset.n_episodes + 1, self.cfg['n'], seed)) env.set_task(task) obs = env.reset() info = env.info reward = 0 total_reward = 0 # Rollout expert policy for _ in range(task.max_steps): act = expert.act(obs, info) episode.append((obs, act, reward, info)) lang_goal = info['lang_goal'] obs, reward, done, info = env.step(act) total_reward += reward print(f'Total Reward: {total_reward:.3f} | Done: {done} | Goal: {lang_goal}') if done: break episode.append((obs, None, reward, info)) return total_reward def simulate_task(self): """ simulate the created task and save demonstrations """ total_cnt = 0. reset_success_cnt = 0. env_success_cnt = 0. seed = 123 self.curr_trials += 1 if p.isConnected(): p.disconnect() if not self.task_creation_pass: print("task creation failure => count as syntax exceptions.") return # Check syntax and compilation-time error try: exec(self.generated_code, globals()) task, dataset, env, expert = self.setup_env() self.syntax_pass_rate += 1 except: to_print = highlight(f"{str(traceback.format_exc())}", PythonLexer(), TerminalFormatter()) save_text(self.cfg['model_output_dir'], self.generated_task_name + '_error', str(traceback.format_exc())) print("========================================================") print("Syntax Exception:", to_print) return try: # Collect environment and collect data from oracle demonstrations. env.generated_code = self.generated_code # Set seeds. episode = [] """ run the new task for one episode """ add_to_txt( self.chat_log, f"================= TRIAL: {self.curr_trials}", with_print=True) np.random.seed(seed) random.seed(seed) print('Oracle demo: {}/{} | Seed: {}'.format(dataset.n_episodes + 1, self.cfg['n'], seed)) env.set_task(task) obs = env.reset() info = env.info reward = 0 total_reward = 0 # Rollout expert policy start_time = time.time() print("start sim") for i in range(task.max_steps): act = expert.act(obs, info) episode.append((obs, act, reward, info)) lang_goal = info['lang_goal'] env.generated_code = self.generated_code yield from env.step(act) obs, reward, done, info = env.cur_obs, env.cur_reward, env.cur_done, env.cur_info total_reward += reward print(f'Total Reward: {total_reward:.3f} | Done: {done} | Goal: {lang_goal}') if done: break end_time = time.time() print("end sim, time used = ", end_time - start_time) yield "Task Generated ==> Asset Generated ==> API Reviewed ==> Error Reviewed ==> Code Generated ==> Simulation Running completed", self.generated_code, env.video_path episode.append((obs, None, reward, info)) # reset_success_cnt += 1 # env_success_cnt += total_reward > 0.99 # # self.runtime_pass_rate += 1 print("Runtime Test Pass!") except: to_print = highlight(f"{str(traceback.format_exc())}", PythonLexer(), TerminalFormatter()) save_text(self.cfg['model_output_dir'], self.generated_task_name + '_error', str(traceback.format_exc())) print("========================================================") print("Runtime Exception:", to_print) self.memory.save_run(self.generated_task)