cortx-labs/bimanual-sandwich-assembly

该数据集是使用LeRobot框架创建的机器人数据集，专注于双手机器人（bi_so_follower）任务。数据集包含55个总剧集和168,768个总帧，用于训练和评估机器人控制算法。数据以parquet文件格式存储，总数据文件大小为100MB，视频文件大小为200MB。特征包括动作（12个浮点数值，对应左右机械臂的12个关节位置）和观察状态（同样12个关节位置），以及来自三台摄像机的视频观察：左顶摄像机、左腕左摄像机和右腕右摄像机，每台摄像机提供480x640分辨率、3通道的RGB视频（AV1编码，30fps）。此外，数据集包含时间戳、帧索引、剧集索引等元数据。所有数据仅用于训练分割（train: 0:55）。数据集适用于机器人学习、模仿学习或强化学习任务，特别是双手机器人操作场景。

This dataset was created using the LeRobot framework and focuses on robotics tasks for a bimanual robot (bi_so_follower). It contains 55 total episodes and 168,768 total frames, designed for training and evaluating robot control algorithms. The data is stored in parquet format, with a total data file size of 100MB and video file size of 200MB. Features include actions (12 floating-point values corresponding to joint positions for left and right arms) and observation states (same 12 joint positions), along with video observations from three cameras: left_top, left_wrist_left, and right_wrist_right, each providing RGB video at 480x640 resolution with 3 channels (AV1 codec, 30fps). Additionally, the dataset includes metadata such as timestamp, frame index, and episode index. All data is allocated for the training split (train: 0:55). The dataset is suitable for robotics learning, imitation learning, or reinforcement learning tasks, particularly in bimanual robot manipulation scenarios.