HyeonseokE/SO101-PlaceSpoon_Ours_40epi
收藏资源简介:
该数据集是一个机器人学习数据集,使用LeRobot创建,专门用于机器人任务(如放置勺子)。数据集包含40个episodes,总计14197帧,针对一个任务(total_tasks: 1)。机器人类型为so101_follower。数据以parquet文件和视频文件形式存储,总数据大小为100MB,视频大小为200MB,帧率为10fps。数据集仅包含训练集(splits: train)。特征包括:机器人状态观测(6个关节位置:肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转、夹爪位置)、动作(6个关节位置,与状态观测相同)、来自顶部和左腕摄像头的图像观测(480x640分辨率,RGB视频,编码为h264)、末端执行器位置(XYZRPY坐标)、夹爪二进制状态、技能自然语言描述、技能验证问题、技能类型、技能进度、技能目标位置(关节和XYZRPY坐标)、子任务自然语言描述、子任务对象名称、子任务目标位置(XYZ坐标),以及时间戳、帧索引、episode索引等元数据。数据集适用于机器人控制、模仿学习或强化学习研究。
This dataset is a robotics learning dataset created using LeRobot, specifically for robot tasks (e.g., placing a spoon). It contains 40 episodes, totaling 14197 frames, for one task (total_tasks: 1). The robot type is so101_follower. Data is stored in parquet files and video files, with a total data size of 100MB and video size of 200MB, at a frame rate of 10fps. The dataset includes only a training set (splits: train). Features include: robot state observation (6 joint positions: shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, gripper), action (6 joint positions, same as state observation), image observations from top and left wrist cameras (480x640 resolution, RGB video, encoded as h264), end-effector position (XYZRPY coordinates), gripper binary state, skill natural language description, skill verification question, skill type, skill progress, skill goal position (joint and XYZRPY coordinates), subtask natural language description, subtask object name, subtask target position (XYZ coordinates), and metadata such as timestamp, frame index, episode index, etc. The dataset is suitable for robotics control, imitation learning, or reinforcement learning research.




