gsethia08/part4-push_pop_toy

该数据集是使用LeRobot创建的机器人手部操作数据集，专为机器人学习任务设计。数据集包含2801个episodes，总计1062318帧，覆盖30个不同的任务。数据以parquet格式存储，视频文件分辨率为1080x1920，帧率为30fps。数据集的特征包括观察数据（如相机图像和机器人状态）、动作数据、语言指令、时间戳、帧索引、episode索引等。其中，观察数据包括来自ego相机的视频（形状为[1080, 1920, 3]）和机器人状态（48维浮点数，表示左右手腕、手指的位置和旋转信息）。动作数据为48维浮点数。此外，还包含相机外参和内参数据。数据集适用于训练和评估机器人控制模型，特别是基于视觉和语言指令的手部操作任务。This dataset is a robotic hand manipulation dataset created using LeRobot, designed for robot learning tasks. It contains 2801 episodes, totaling 1062318 frames, and covers 30 different tasks. The data is stored in parquet format, with video files at a resolution of 1080x1920 and a frame rate of 30fps. The dataset features include observation data (such as camera images and robot state), action data, language instructions, timestamps, frame indices, episode indices, and more. Specifically, observation data includes video from an ego camera (shape [1080, 1920, 3]) and robot state (48-dimensional float values representing positions and rotations of left and right wrists and fingers). Action data is 48-dimensional float values. Additionally, camera extrinsic and intrinsic parameters are included. The dataset is suitable for training and evaluating robot control models, particularly for vision- and language-instructed hand manipulation tasks.