hangX/lerobot
1
0
Fork 0
forked from tangger/lerobot
Code Pull Requests Activity

enabled saving data in degrees

Browse Source
This commit is contained in:
Michel Aractingi
2024-10-22 01:19:11 +02:00
parent 0ef2397029
commit 8b17416fc7
1 changed files with 5 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
lerobot/scripts/control_sim_robot.py
View File

@@ -194,7 +194,7 @@ def read_commands_from_leader(robot: Robot, queue: multiprocessing.Queue, fps: i
start_pos = np.array(robot.leader_arms.main.calibration['start_pos']) start_pos = np.array(robot.leader_arms.main.calibration['start_pos'])
axis_directions = np.array(axis_directions) axis_directions = np.array(axis_directions)
offsets = np.array(offsets) * np.pi offsets = np.array(offsets) * np.pi
counts_to_radians = np.pi * 2. / 4096 counts_to_radians = 2.0 * np.pi / 4096
if stop_flag is None: if stop_flag is None:
stop_flag = multiprocessing.Value('b', False) stop_flag = multiprocessing.Value('b', False)
@@ -391,14 +391,15 @@ def record(
state_obs = [] state_obs = []
for key in state_keys: for key in state_keys:
state_obs.append(torch.from_numpy(observation[key])) state_obs.append(torch.from_numpy(observation[key]))
ep_dict['observation.state'].append(torch.hstack(state_obs)) ep_dict['observation.state'].append(torch.hstack(state_obs) * 180.0 / np.pi)
# Advance the sim environment # Advance the sim environment
if len(action.shape) == 1: if len(action.shape) == 1:
action = np.expand_dims(action, 0) action = np.expand_dims(action, 0)
observation, reward, _, _ , info = env.step(action) observation, reward, _, _ , info = env.step(action)
ep_dict['action'].append(torch.from_numpy(action)) ep_dict['action'].append(torch.from_numpy(action) * 180.0 / np.pi)
ep_dict['reward'].append(torch.tensor(reward)) ep_dict['reward'].append(torch.tensor(reward))
print(reward)
frame_index += 1 frame_index += 1
@@ -590,7 +591,7 @@ def replay(env, episodes: list, fps: int | None = None, root="data", repo_id="le
action = items[idx]["action"] action = items[idx]["action"]
env.step(action.unsqueeze(0).numpy()) env.step(action.unsqueeze(0).numpy() * np.pi / 180.0)
dt_s = time.perf_counter() - start_episode_t dt_s = time.perf_counter() - start_episode_t
busy_wait(1 / fps - dt_s) busy_wait(1 / fps - dt_s)