CoRL2026-CSI/SO101-cap_multitask_7tasks_700epi_10fps

该数据集是一个用于机器人技术的多任务数据集，使用LeRobot框架创建。它包含700个episodes，总计362,160帧，覆盖7个不同的任务。数据以10fps的视频格式存储，包括顶部摄像头和左腕摄像头的观察图像（分辨率480x640，RGB三通道）。数据集提供了丰富的特征，如机器人关节状态（6个自由度）、动作指令、末端执行器位置（XYZRPY坐标）、夹爪状态、自然语言技能描述、验证问题、技能类型和进度、目标位置（关节和笛卡尔坐标）、子任务信息（自然语言描述、对象名称、目标位置）以及时间戳和索引信息。数据格式为Parquet文件，总数据文件大小为100MB，视频文件大小为200MB。该数据集适用于机器人学习、策略编码和多任务强化学习等研究。This dataset is a multi-task dataset for robotics, created using the LeRobot framework. It contains 700 episodes, totaling 362,160 frames, covering 7 different tasks. The data is stored in video format at 10fps, including observation images from a top camera and a left wrist camera (resolution 480x640, RGB three channels). The dataset provides rich features such as robot joint states (6 degrees of freedom), action commands, end-effector positions (XYZRPY coordinates), gripper status, natural language skill descriptions, verification questions, skill types and progress, goal positions (joint and Cartesian coordinates), subtask information (natural language descriptions, object names, target positions), as well as timestamps and index information. The data format is Parquet files, with a total data file size of 100MB and video file size of 200MB. This dataset is suitable for research in robot learning, policy coding, and multi-task reinforcement learning.