HyeonseokE/SO101-CloseDrawer_Ours_20epi
收藏资源简介:
该数据集是通过LeRobot平台创建的机器人学数据集,专注于关闭抽屉任务。数据集包含20个训练片段,总计5320帧数据,帧率为10fps。数据以Parquet格式存储,并包含视频文件。特征包括机器人关节状态(6维浮点数)、动作(6维浮点数)、顶部和左腕摄像头图像(480x640 RGB视频)、末端执行器位置(6维浮点数)、夹爪二进制状态、技能自然语言描述、验证问题、技能类型、进度、目标位置(关节和笛卡尔坐标)、子任务自然语言描述、对象名称、目标位置(3维浮点数)、时间戳、帧索引、片段索引、索引和任务索引。数据集适用于机器人控制、模仿学习或强化学习任务。
This dataset is a robotics dataset created using the LeRobot platform, focusing on the task of closing a drawer. It includes 20 training episodes with a total of 5320 frames at a frame rate of 10fps. The data is stored in Parquet format and includes video files. Features encompass robot joint states (6-dimensional floats), actions (6-dimensional floats), images from top and left wrist cameras (480x640 RGB video), end-effector positions (6-dimensional floats), gripper binary state, natural language descriptions of skills, verification questions, skill types, progress, goal positions (joint and Cartesian coordinates), natural language descriptions of subtasks, object names, target positions (3-dimensional floats), timestamps, frame indices, episode indices, indices, and task indices. The dataset is suitable for robotics control, imitation learning, or reinforcement learning tasks.




