RLP-ETH/RobotLearningData_v3-1

该数据集是一个用于机器人学习的演示数据集，通过LeRobot创建。它包含673个episodes、118,857帧和29个任务，数据以30fps的帧率采集。数据集的特征包括机器人的动作（如肩部、肘部、腕部和抓手的关节位置）、状态观察（与动作相同的关节位置）、来自前置摄像头的图像观察（分辨率为480x640的RGB视频），以及时间戳、帧索引、episode索引、任务索引、完成标志等元数据。数据以parquet格式存储，视频以mp4格式存储，总数据大小约为300MB。机器人类型为so_follower，适用于训练和评估机器人控制算法。This dataset is a demonstration dataset for robot learning, created using LeRobot. It contains 673 episodes, 118,857 frames, and 29 tasks, with data collected at 30fps. The dataset features include robot actions (e.g., joint positions for shoulder, elbow, wrist, and gripper), state observations (same joint positions as actions), image observations from a front camera (RGB video with 480x640 resolution), and metadata such as timestamp, frame index, episode index, task index, and done flag. The data is stored in parquet format, videos in mp4 format, with a total size of approximately 300MB. The robot type is so_follower, suitable for training and evaluating robot control algorithms.