stevenworkspace/eval_take_act_8
收藏资源简介:
该数据集使用LeRobot创建,专为移动AI机器人设计。数据集包含一个完整训练集(共1个episode,1049帧,30fps),数据涵盖机器人动作控制、状态观测和多视角视觉信息。具体特征包括:16维动作向量(控制左右机械臂的关节位置和速度)、16维状态观测向量(反映机器人当前关节状态)、三个摄像头视频流(高分辨率摄像头、左腕摄像头、右腕摄像头,均为480x640分辨率RGB视频)。数据集格式为parquet,适用于机器人学习任务,如模仿学习或强化学习。
This dataset was created using LeRobot and is designed for mobile AI robots. It contains a complete training set (1 episode, 1049 frames at 30fps), encompassing robot action control, state observations, and multi-view visual information. Specific features include: 16-dimensional action vectors (controlling joint positions and velocities of left/right robotic arms), 16-dimensional state observation vectors (reflecting current robot joint states), and three camera video streams (high-resolution camera, left wrist camera, right wrist camera, all 480x640 resolution RGB videos). The dataset is in parquet format and suitable for robot learning tasks such as imitation learning or reinforcement learning.




