eval_2026-2-4-15

该数据集是一个机器人操作数据集，使用LeRobot创建。包含1个任务场景的2174帧数据，帧率为30fps。数据集记录了SO101型跟随机器人的6维关节位置动作(肩部平移、肩部抬升、肘部弯曲、腕部弯曲、腕部旋转和夹持器位置)和对应的6维状态观测。同时采集了三个视角(手腕、侧面和头部)的480x640分辨率RGB视频数据。数据以parquet格式存储，总数据量约100MB，视频数据约200MB。数据集主要用于机器人学习和控制研究。This is a robotic manipulation dataset created using LeRobot. It contains 2174 frames of data from a single task scenario, with a frame rate of 30 fps. The dataset records the 6-dimensional joint position actions (shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position) and corresponding 6-dimensional state observations of the SO101 follow robot. Additionally, RGB video footage with a resolution of 480×640 was captured from three perspectives: wrist, side, and head. The structured data including joint actions and state observations is stored in Parquet format, with a total size of approximately 100 MB, while the video data amounts to about 200 MB. This dataset is primarily designed for robotic learning and control research.