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

set boundary condition to None at reset() & end replanning if env is done

Browse Source
This commit is contained in:
Hongyi Zhou 2022-11-27 01:25:31 +01:00
parent 42863cee1c
commit bf0de5f54d
2 changed files with 7 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
fancy_gym/black_box/black_box_wrapper.py
View File

@ -161,9 +161,7 @@ class BlackBoxWrapper(gym.ObservationWrapper):
self.plan_steps += 1 self.plan_steps += 1
for t, (pos, vel) in enumerate(zip(position, velocity)): for t, (pos, vel) in enumerate(zip(position, velocity)):
current_pos = self.current_pos step_action = self.tracking_controller.get_action(pos, vel, self.current_pos, self.current_vel)
current_vel = self.current_vel
step_action = self.tracking_controller.get_action(pos, vel, current_pos, current_vel)
c_action = np.clip(step_action, self.env.action_space.low, self.env.action_space.high) c_action = np.clip(step_action, self.env.action_space.low, self.env.action_space.high)
obs, c_reward, done, info = self.env.step(c_action) obs, c_reward, done, info = self.env.step(c_action)
rewards[t] = c_reward rewards[t] = c_reward
@ -180,10 +178,10 @@ class BlackBoxWrapper(gym.ObservationWrapper):
if self.render_kwargs: if self.render_kwargs:
self.env.render(**self.render_kwargs) self.env.render(**self.render_kwargs)
if done or self.replanning_schedule(current_pos, current_vel, obs, c_action, if done or self.replanning_schedule(self.current_pos, self.current_vel, obs, c_action,
t + 1 + self.current_traj_steps): t + 1 + self.current_traj_steps):
if self.max_planning_times is not None and self.plan_steps >= self.max_planning_times: if not done and self.max_planning_times is not None and self.plan_steps >= self.max_planning_times:
continue continue
self.condition_pos = pos if self.condition_on_desired else None self.condition_pos = pos if self.condition_on_desired else None
@ -214,4 +212,6 @@ class BlackBoxWrapper(gym.ObservationWrapper):
self.current_traj_steps = 0 self.current_traj_steps = 0
self.plan_steps = 0 self.plan_steps = 0
self.traj_gen.reset() self.traj_gen.reset()
self.condition_vel = None
self.condition_pos = None
return super(BlackBoxWrapper, self).reset() return super(BlackBoxWrapper, self).reset()

4
fancy_gym/envs/__init__.py
View File

@ -485,7 +485,7 @@ for _v in _versions:
for _v in _versions: for _v in _versions:
_name = _v.split("-") _name = _v.split("-")
_env_id = f'{_name[0]}ReplanProDMP-{_name[1]}' _env_id = f'{_name[0]}ProDMP-{_name[1]}'
kwargs_dict_box_pushing_prodmp = deepcopy(DEFAULT_BB_DICT_ProDMP) kwargs_dict_box_pushing_prodmp = deepcopy(DEFAULT_BB_DICT_ProDMP)
kwargs_dict_box_pushing_prodmp['wrappers'].append(mujoco.box_pushing.MPWrapper) kwargs_dict_box_pushing_prodmp['wrappers'].append(mujoco.box_pushing.MPWrapper)
kwargs_dict_box_pushing_prodmp['name'] = _v kwargs_dict_box_pushing_prodmp['name'] = _v
@ -498,7 +498,7 @@ for _v in _versions:
kwargs_dict_box_pushing_prodmp['basis_generator_kwargs']['num_basis'] = 4 kwargs_dict_box_pushing_prodmp['basis_generator_kwargs']['num_basis'] = 4
kwargs_dict_box_pushing_prodmp['basis_generator_kwargs']['basis_bandwidth_factor'] = 3 kwargs_dict_box_pushing_prodmp['basis_generator_kwargs']['basis_bandwidth_factor'] = 3
kwargs_dict_box_pushing_prodmp['phase_generator_kwargs']['alpha_phase'] = 3 kwargs_dict_box_pushing_prodmp['phase_generator_kwargs']['alpha_phase'] = 3
kwargs_dict_box_pushing_prodmp['black_box_kwargs']['max_planning_times'] = 2 kwargs_dict_box_pushing_prodmp['black_box_kwargs']['max_planning_times'] = 4
kwargs_dict_box_pushing_prodmp['black_box_kwargs']['replanning_schedule'] = lambda pos, vel, obs, action, t : t % 25 == 0 kwargs_dict_box_pushing_prodmp['black_box_kwargs']['replanning_schedule'] = lambda pos, vel, obs, action, t : t % 25 == 0
kwargs_dict_box_pushing_prodmp['black_box_kwargs']['condition_on_desired'] = True kwargs_dict_box_pushing_prodmp['black_box_kwargs']['condition_on_desired'] = True
register( register(