eval_act_grab_release_07_01

该数据集使用LeRobot创建，包含5个训练集，总帧数为1616帧，采用so_follower机器人类型。数据集结构包括动作特征（如肩部、肘部、腕部和夹爪位置）、状态观测（与动作特征相同）、手机图像（640x480分辨率）和手腕图像（480x640分辨率），所有图像均为彩色视频格式，帧率为25fps。数据以parquet文件存储，视频以mp4格式存储，总数据文件大小为100MB，视频文件大小为200MB。

This dataset is created using LeRobot and contains 5 training sets with a total of 1616 frames, employing the so_follower robot type. The dataset structure includes action features (such as shoulder, elbow, wrist, and gripper positions), state observations (same as action features), phone images (640x480 resolution), and wrist images (480x640 resolution), with all images in color video format at a frame rate of 25fps. Data is stored in parquet files, videos are stored in mp4 format, with a total data file size of 100MB and video file size of 200MB.