File size: 2,333 Bytes
912c7e2 | 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 63 | import hydra
import numpy as np
from tqdm import tqdm
from omegaconf import OmegaConf
from rlbench.backend.observation import Observation
from pfp import DATA_DIRS, set_seeds
from pfp.envs.rlbench_env import RLBenchEnv
from pfp.data.replay_buffer import RobotReplayBuffer
from pfp.common.visualization import RerunViewer as RV
# For valid, call it with: --config-name=collect_demos_valid
# To actually save the data, remember to call it with: save_data=True
@hydra.main(version_base=None, config_path="../conf", config_name="collect_demos_train")
def main(cfg: OmegaConf):
set_seeds(cfg.seed)
if not OmegaConf.has_resolver("eval"):
OmegaConf.register_new_resolver("eval", eval)
print(OmegaConf.to_yaml(cfg))
assert cfg.mode in ["train", "valid"]
if cfg.env_config.vis:
RV("pfp_collect_demos")
env = RLBenchEnv(use_pc_color=True, **cfg.env_config)
if cfg.save_data:
data_path = DATA_DIRS.PFP / cfg.env_config.task_name / cfg.mode
if data_path.is_dir():
print(f"ERROR: Data path {data_path} already exists! Exiting...")
return
replay_buffer = RobotReplayBuffer.create_from_path(data_path, mode="a")
for _ in tqdm(range(cfg.num_episodes)):
data_history = list()
demo = env.task.get_demos(1, live_demos=True)[0]
observations: list[Observation] = demo._observations
for obs in observations:
robot_state = env.get_robot_state(obs)
images = env.get_images(obs)
pcd = env.get_pcd(obs)
pcd_xyz = np.asarray(pcd.points)
pcd_color = np.asarray(pcd.colors)
data_history.append(
{
"pcd_xyz": pcd_xyz.astype(np.float32),
"pcd_color": (pcd_color * 255).astype(np.uint8),
"robot_state": robot_state.astype(np.float32),
"images": images,
}
)
env.vis_step(robot_state, np.concatenate((pcd_xyz, pcd_color), axis=-1))
if cfg.save_data:
replay_buffer.add_episode_from_list(data_history, compressors="disk")
print(f"Saved episode with {len(data_history)} steps to disk.")
# while True:
# env.step(robot_state)
return
if __name__ == "__main__":
main()
|