HyeonseokE/SO101-CloseDrawer_Ours_60epi
收藏资源简介:
该数据集由LeRobot创建,用于机器人学习任务,具体针对SO101机器人关闭抽屉的操作。数据集包含60个episodes,总计15666帧,数据以parquet文件和视频文件形式存储。特征包括机器人关节状态观测(6维浮点数)、动作指令(6维浮点数)、顶部和左腕摄像头的图像观测(480x640分辨率、3通道、10fps视频)、末端执行器位置(6维浮点数表示XYZ和欧拉角)、抓取器二进制状态、技能和子任务的自然语言描述、技能类型、进度、目标位置(关节和笛卡尔空间)、子任务对象名称和目标位置,以及时间戳、帧索引、episode索引等元数据。数据分割为训练集(全部60个episodes),适用于机器人模仿学习或强化学习研究。
This dataset was created using LeRobot for robotics tasks, specifically focusing on the operation of closing a drawer with an SO101 robot. It contains 60 episodes with a total of 15,666 frames, stored in parquet and video files. Features include robot joint state observations (6-dimensional float32), action commands (6-dimensional float32), image observations from top and left wrist cameras (480x640 resolution, 3-channel, 10fps video), end-effector position (6-dimensional float32 for XYZ and Euler angles), gripper binary state, natural language descriptions for skills and subtasks, skill type, progress, goal positions (joint and Cartesian space), subtask object names and target positions, along with metadata such as timestamp, frame index, and episode index. The data is split into a training set (all 60 episodes), suitable for research in robot imitation learning or reinforcement learning.




