VarunGiridhar3/libero40_libero_10_v2_play

该数据集是使用LeRobot框架创建的机器人任务数据集，专门针对panda机器人设计。数据集包含200个episodes、104000帧数据，覆盖10个不同任务。数据以parquet文件格式存储，总数据文件大小为100MB，视频文件大小为200MB（但视频路径为空，表示可能未包含视频文件）。帧率为10fps。数据集特征包括：动作（7维浮点数组）、观察状态（8维浮点数组，包含末端执行器位置、轴角表示和夹爪状态）、两个图像观察（均为256x256x3的图像）、时间戳、帧索引、episode索引、任务索引等。数据集适用于机器人控制、强化学习或模仿学习等研究，标签包括LeRobot、libero和play_data，表明它可能用于自由操作任务或游戏数据收集。This dataset was created using the LeRobot framework and is designed for robotics tasks, specifically for the panda robot. It contains 200 episodes, 104,000 frames, and covers 10 different tasks. The data is stored in parquet format, with a total data file size of 100MB and video file size of 200MB (though the video path is null, indicating no video files are included). The frame rate is 10fps. Features include: action (7-dimensional float32 array), observation state (8-dimensional float32 array with end-effector position, axis-angle representation, and gripper states), two image observations (both 256x256x3 images), timestamp, frame index, episode index, task index, etc. The dataset is suitable for research in robot control, reinforcement learning, or imitation learning, and is tagged with LeRobot, libero, and play_data, suggesting it may be used for libre manipulation tasks or play data collection.