dodox/fancy_gym
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

viapoint reacher reward bug fix

Browse Source
This commit is contained in:
Maximilian Huettenrauch 2021-02-19 16:17:55 +01:00
parent 1e3f036478
commit 60e1673ee1
3 changed files with 15 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
alr_envs/classic_control/utils.py
View File

@ -23,11 +23,12 @@ def make_viapointreacher_env(rank, seed=0):
num_dof=5, num_dof=5,
num_basis=5, num_basis=5,
duration=2, duration=2,
alpha_phase=2, alpha_phase=2.5,
dt=_env.dt, dt=_env.dt,
start_pos=_env.start_pos, start_pos=_env.start_pos,
learn_goal=False, learn_goal=False,
policy_type="velocity") policy_type="velocity",
weights_scale=50)
_env.seed(seed + rank) _env.seed(seed + rank)
return _env return _env

2
alr_envs/classic_control/viapoint_reacher.py
View File

@ -83,7 +83,7 @@ class ViaPointReacher(gym.Env):
if not self._is_collided: if not self._is_collided:
if self._steps == 100: if self._steps == 100:
dist_reward = np.linalg.norm(self.end_effector - self.via_point) dist_reward = np.linalg.norm(self.end_effector - self.via_point)
if self._steps == 200: if self._steps == 199:
dist_reward = np.linalg.norm(self.end_effector - self.goal_point) dist_reward = np.linalg.norm(self.end_effector - self.goal_point)
reward = - dist_reward ** 2 reward = - dist_reward ** 2

13
dmp_env_wrapper_example.py
View File

@ -14,10 +14,19 @@ if __name__ == "__main__":
test_env = make_viapointreacher_env(0)() test_env = make_viapointreacher_env(0)()
params = np.random.randn(n_samples, dim) # params = np.random.randn(n_samples, dim)
params = np.array([ 217.54494933, -1.85169983, 24.08414447, 42.23816868,
23.32071702, 7.60780651, -31.74777741, 265.50634253,
463.43822562, 245.93948374, -272.64003621, -45.24999553,
503.21185823, 809.17742517, 393.12387021, -196.54196471,
6.79327307, 374.82429078, 552.4119579 , 197.3963343 ,
243.87357056, -39.56041541, -616.93957463, -710.0772516 ,
-414.21769789])
# params = np.hstack([50 * np.random.randn(n_samples, 25), np.tile(np.array([np.pi/2, -np.pi/4, -np.pi/4, -np.pi/4, -np.pi/4]), [n_samples, 1])]) # params = np.hstack([50 * np.random.randn(n_samples, 25), np.tile(np.array([np.pi/2, -np.pi/4, -np.pi/4, -np.pi/4, -np.pi/4]), [n_samples, 1])])
test_env.rollout(params, render=True) rew, info = test_env.rollout(params, render=True)
print(rew)
# out = env(params) # out = env(params)
# print(out) # print(out)