csacan/SO101-eval2-brg-25prompts

该数据集是由LeRobot创建的机器人操作数据集，专门用于机器人技术研究。数据集包含250个完整操作片段，总计112015帧数据，涵盖25个不同任务。数据采集自so_follower类型机器人，以30fps的帧率记录。数据集包含机械臂的6个关节位置信息（肩部平移、肩部升降、肘部弯曲、腕部弯曲、腕部旋转和夹爪位置），以及480x640分辨率的RGB前端摄像头视频流。所有数据以parquet格式存储，总数据量约100MB，视频文件约200MB。数据集仅包含训练集划分，适用于机器人模仿学习、行为克隆等任务。This dataset is a robotics manipulation dataset created using LeRobot, designed for robotics research. It contains 250 complete episodes with a total of 112,015 frames, covering 25 distinct tasks. Data was collected from a so_follower type robot at 30fps. The dataset includes 6-degree-of-freedom joint position information for the robotic arm (shoulder pan, shoulder lift, elbow flex, wrist flex, wrist roll, and gripper positions), along with front-facing RGB camera video streams at 480x640 resolution. All data is stored in parquet format, with approximately 100MB of data files and 200MB of video files. The dataset includes only a training split and is suitable for tasks such as robotic imitation learning and behavior cloning.