Lakesenberg/Seeed_Hackathon_egg_2

该数据集由LeRobot创建，用于机器人学任务，具体针对Seeed B601 DM Follower机器人。数据集包含16个episodes，总计11984个frames，数据以parquet文件格式存储，视频以mp4格式存储。特征包括：7维浮点动作向量（对应肩部平移、肩部提升、肘部弯曲、腕部弯曲、腕部偏航、腕部滚动和夹爪位置）；7维浮点状态观测向量（与动作相同）；两个视角的图像观测：side视角（880x640分辨率，3通道RGB视频）和hand视角（480x640分辨率，3通道RGB视频），视频帧率为30fps；以及时间戳、帧索引、episode索引、任务索引等元数据。数据集仅包含训练集，数据分块大小为1000，总数据文件大小约100MB，视频文件大小约200MB。

This dataset was created by LeRobot for robotics tasks, specifically for the Seeed B601 DM Follower robot. It contains 16 episodes with a total of 11984 frames, stored in parquet format for data and mp4 for videos. Features include: a 7-dimensional float32 action vector (corresponding to shoulder pan, shoulder lift, elbow flex, wrist flex, wrist yaw, wrist roll, and gripper positions); a 7-dimensional float32 state observation vector (same as actions); image observations from two perspectives: side view (880x640 resolution, 3-channel RGB video) and hand view (480x640 resolution, 3-channel RGB video), with video frame rate of 30fps; and metadata such as timestamp, frame index, episode index, and task index. The dataset includes only a training set, with a chunk size of 1000, total data file size of approximately 100MB, and video file size of approximately 200MB.