stevenworkspace/eval_take_17
收藏资源简介:
该数据集是一个用于机器人学习的移动AI机器人数据集,包含1个任务和1个episode,总帧数为1029。数据以30 fps的帧率采集,存储为parquet格式,视频文件使用av1编解码器。数据集特征包括16维的动作和状态观测(如左右关节位置和速度),以及三个摄像头(cam_high、cam_left_wrist、cam_right_wrist)采集的480x640 RGB视频观测。数据集结构支持分块处理,每个chunk大小为1000帧,总数据文件大小为100 MB,视频文件大小为200 MB。数据集适用于机器人控制、模仿学习等任务,基于Apache 2.0许可证发布。
This dataset is a mobile AI robot dataset for robotics learning, containing 1 task and 1 episode with a total of 1029 frames. Data is collected at 30 fps, stored in parquet format, and video files use the av1 codec. Dataset features include 16-dimensional actions and state observations (e.g., left and right joint positions and velocities), as well as video observations from three cameras (cam_high, cam_left_wrist, cam_right_wrist) at 480x640 RGB resolution. The dataset structure supports chunked processing with each chunk size of 1000 frames, total data file size of 100 MB, and video file size of 200 MB. It is suitable for tasks such as robot control and imitation learning, released under the Apache 2.0 license.




