L7-Robotics/so101_mmc_3cam_red_cube_white_bowl_v2
收藏资源简介:
该数据集是使用LeRobot创建的机器人数据集,专为机器人任务设计。数据集包含10个episodes,总计4727帧,帧率为20 FPS。数据以Parquet文件格式存储,视频文件为MP4格式。数据集的特征包括:动作数据(action),包含6个浮点数值,表示机器人关节位置(如肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置);观察状态(observation.state),同样包含6个浮点数值,表示机器人状态;以及三个摄像头的图像观察(observation.images.rear_top、observation.images.right_side、observation.images.wrist),每个视频分辨率为480x640,3通道,使用AV1编解码器。此外,数据集还包括时间戳、帧索引、episode索引、索引和任务索引等元数据。机器人类型为so_follower,数据分割为训练集(0:10)。数据集总大小为100 MB(数据文件)和200 MB(视频文件),适用于机器人学习和控制任务。
This dataset is a robotics dataset created using LeRobot, designed for robotics tasks. It contains 10 episodes with a total of 4727 frames at a frame rate of 20 FPS. The data is stored in Parquet file format, with video files in MP4 format. The dataset features include: action data, consisting of 6 float32 values representing robot joint positions (e.g., shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions); observation state, also with 6 float32 values for robot state; and image observations from three cameras (rear_top, right_side, wrist), each video with a resolution of 480x640, 3 channels, using AV1 codec. Additionally, the dataset includes metadata such as timestamp, frame index, episode index, index, and task index. The robot type is so_follower, with a split for training set (0:10). The total dataset size is 100 MB for data files and 200 MB for video files, suitable for robotics learning and control tasks.




