vla-grab-dataset-v1-4

该数据集是通过LeRobot创建的机器人相关数据集，包含20个episodes，13662帧，30fps的视频数据。数据集结构包括动作、观察状态、图像等多种特征。具体来说，动作和观察状态都包含6个浮点型数据，分别对应机器人的各个关节位置。图像数据为480x640分辨率的彩色视频。

This is a robotics-related dataset created via LeRobot, which contains 20 episodes, 13662 frames of 30fps video data. The dataset structure includes multiple features such as actions, observation states, and image data. Specifically, both the actions and observation states consist of 6 floating-point values, which respectively correspond to the joint positions of the robot. The image data is color video with a resolution of 480×640.