57 lines
1.6 KiB
Python
57 lines
1.6 KiB
Python
import numpy as np
|
|
|
|
from gymnasium import utils
|
|
from gymnasium.envs.mujoco import MuJocoPyEnv
|
|
from gymnasium.spaces import Box
|
|
|
|
|
|
class InvertedPendulumEnv(MuJocoPyEnv, utils.EzPickle):
|
|
metadata = {
|
|
"render_modes": [
|
|
"human",
|
|
"rgb_array",
|
|
"depth_array",
|
|
],
|
|
"render_fps": 25,
|
|
}
|
|
|
|
def __init__(self, **kwargs):
|
|
utils.EzPickle.__init__(self, **kwargs)
|
|
observation_space = Box(low=-np.inf, high=np.inf, shape=(4,), dtype=np.float64)
|
|
MuJocoPyEnv.__init__(
|
|
self,
|
|
"inverted_pendulum.xml",
|
|
2,
|
|
observation_space=observation_space,
|
|
**kwargs,
|
|
)
|
|
|
|
def step(self, a):
|
|
reward = 1.0
|
|
self.do_simulation(a, self.frame_skip)
|
|
|
|
ob = self._get_obs()
|
|
terminated = bool(not np.isfinite(ob).all() or (np.abs(ob[1]) > 0.2))
|
|
|
|
if self.render_mode == "human":
|
|
self.render()
|
|
return ob, reward, terminated, False, {}
|
|
|
|
def reset_model(self):
|
|
qpos = self.init_qpos + self.np_random.uniform(
|
|
size=self.model.nq, low=-0.01, high=0.01
|
|
)
|
|
qvel = self.init_qvel + self.np_random.uniform(
|
|
size=self.model.nv, low=-0.01, high=0.01
|
|
)
|
|
self.set_state(qpos, qvel)
|
|
return self._get_obs()
|
|
|
|
def _get_obs(self):
|
|
return np.concatenate([self.sim.data.qpos, self.sim.data.qvel]).ravel()
|
|
|
|
def viewer_setup(self):
|
|
assert self.viewer is not None
|
|
self.viewer.cam.trackbodyid = 0
|
|
self.viewer.cam.distance = self.model.stat.extent
|