yoohoolala/depth_ai02

该数据集是由LeRobot创建的机器人学习数据集，包含机器人动作、观测状态和图像等特征。数据集结构包括6个浮点型动作特征（肩关节、肘关节、手腕和夹持器的位置）、6个浮点型观测状态特征、720x1280分辨率的RGB图像视频数据（30fps）、时间戳、帧索引、情景索引等。总共有2个情景、598帧数据和1个任务，数据以parquet格式存储。This is a robotic learning dataset developed by LeRobot, which includes features such as robot actions, observation states, and images. The dataset structure consists of 6 floating-point action features (positions of the shoulder joint, elbow joint, wrist, and gripper), 6 floating-point observation state features, RGB image video data with a resolution of 720×1280 (30 fps), timestamps, frame indices, episode indices, and more. In total, the dataset contains 2 episodes, 598 frames, and 1 task, with all data stored in Parquet format.