Spaces:
Running
Running
File size: 1,897 Bytes
375a1cf |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 |
import numpy as np
from gym import utils
from gym.envs.mujoco import MuJocoPyEnv
from gym.spaces import Box
class Walker2dEnv(MuJocoPyEnv, utils.EzPickle):
metadata = {
"render_modes": [
"human",
"rgb_array",
"depth_array",
],
"render_fps": 125,
}
def __init__(self, **kwargs):
observation_space = Box(low=-np.inf, high=np.inf, shape=(17,), dtype=np.float64)
MuJocoPyEnv.__init__(
self, "walker2d.xml", 4, observation_space=observation_space, **kwargs
)
utils.EzPickle.__init__(self, **kwargs)
def step(self, a):
posbefore = self.sim.data.qpos[0]
self.do_simulation(a, self.frame_skip)
posafter, height, ang = self.sim.data.qpos[0:3]
alive_bonus = 1.0
reward = (posafter - posbefore) / self.dt
reward += alive_bonus
reward -= 1e-3 * np.square(a).sum()
terminated = not (height > 0.8 and height < 2.0 and ang > -1.0 and ang < 1.0)
ob = self._get_obs()
if self.render_mode == "human":
self.render()
return ob, reward, terminated, False, {}
def _get_obs(self):
qpos = self.sim.data.qpos
qvel = self.sim.data.qvel
return np.concatenate([qpos[1:], np.clip(qvel, -10, 10)]).ravel()
def reset_model(self):
self.set_state(
self.init_qpos
+ self.np_random.uniform(low=-0.005, high=0.005, size=self.model.nq),
self.init_qvel
+ self.np_random.uniform(low=-0.005, high=0.005, size=self.model.nv),
)
return self._get_obs()
def viewer_setup(self):
assert self.viewer is not None
self.viewer.cam.trackbodyid = 2
self.viewer.cam.distance = self.model.stat.extent * 0.5
self.viewer.cam.lookat[2] = 1.15
self.viewer.cam.elevation = -20
|