Lakesenberg/Seeed_rebot_lianxucaiji_18_lian3_merged

该数据集是一个机器人控制数据集，使用LeRobot平台创建，专门针对seeed_b601_dm_follower机器人类型。数据集包含28个episodes，总计50285帧，帧率为30fps，仅用于训练任务。数据以parquet格式存储，总大小为100MB，同时包含视频文件，总大小为200MB。特征包括动作数据（7个关节位置：shoulder_pan.pos、shoulder_lift.pos、elbow_flex.pos、wrist_flex.pos、wrist_yaw.pos、wrist_roll.pos、gripper.pos）、观察状态（同样为7个关节位置）、来自侧视和手部摄像头的图像观察（分辨率分别为880x640和480x640，RGB三通道）、时间戳、帧索引、episode索引、索引和任务索引。数据集结构支持机器人学习和控制任务，适用于强化学习或模仿学习应用。

This dataset is a robotics control dataset created using the LeRobot platform, specifically for the seeed_b601_dm_follower robot type. It contains 28 episodes, totaling 50,285 frames at 30fps, and is intended for training tasks only. Data is stored in parquet format with a total size of 100MB, and includes video files with a total size of 200MB. Features include action data (7 joint positions: shoulder_pan.pos, shoulder_lift.pos, elbow_flex.pos, wrist_flex.pos, wrist_yaw.pos, wrist_roll.pos, gripper.pos), observation state (same 7 joint positions), image observations from side and hand cameras (resolutions 880x640 and 480x640 respectively, with 3 RGB channels), timestamp, frame index, episode index, index, and task index. The dataset structure supports robot learning and control tasks, suitable for reinforcement learning or imitation learning applications.