so100_battery-trajectory

该数据集是一个机器人轨迹数据集，使用LeRobot工具创建，专门针对so100机器人类型。数据集包含50个episodes，总计24263帧数据，涉及一个任务。数据以parquet文件格式存储，视频文件以mp4格式存储，帧率为30fps。特征包括动作（6维浮点数组，对应机器人的主要关节控制，如肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转和夹持器）、观察状态（6维浮点数组，与动作类似）、时间戳、帧索引、episode索引、任务索引，以及来自多个摄像头的图像观察：MAIN_CAM和SEC_CAM（均为480x640x3的视频，30fps，使用av1编解码器）和轨迹图像（480x640x3的静态图像）。数据集主要用于机器人学习和控制任务，支持训练和评估算法。

This dataset is a robotic trajectory dataset developed using the LeRobot toolkit, specifically tailored for the so100 robot platform. It contains 50 episodes, totaling 24,263 frames, and focuses on one single task. The structured data is stored in Parquet file format, while the video files are saved in MP4 format with a frame rate of 30 fps. The dataset includes the following features: actions (6-dimensional floating-point arrays corresponding to the robot's main joint control commands, namely shoulder translation, shoulder elevation, elbow flexion, wrist flexion, wrist rotation, and the gripper), observation states (6-dimensional floating-point arrays similar to the action space), timestamps, frame indices, episode indices, task indices, as well as image observations from multiple cameras: MAIN_CAM and SEC_CAM (both 480x640x3 resolution videos at 30 fps, using the AV1 codec), and trajectory images (480x640x3 static images). This dataset is mainly used for robotic learning and control tasks, supporting algorithm training and evaluation.