Lisette1231/20251013so101

该数据集是一个用于机器人任务的数据集，包含了50个剧集，共34050帧。数据集由Parquet文件组成，每个文件包含1000个数据块。数据集的结构包括动作、观察状态、手眼图像、固定图像、时间戳、帧索引、剧集索引、索引和任务索引等信息。每个动作和观察状态包含肩部、肘部、手腕和夹子的位置信息。手眼图像和固定图像的尺寸为480x640，使用AV1编解码器，每秒30帧，没有音频。

This dataset is for robotics tasks, containing 50 episodes with a total of 34,050 frames. The dataset is composed of Parquet files, each containing 1,000 data chunks. The structure of the dataset includes action, observation state, hand-eye images, fixed images, timestamp, frame index, episode index, index, and task index. Each action and observation state includes position information for the shoulder, elbow, wrist, and gripper. The hand-eye and fixed images are 480x640 in size, encoded with AV1 codec, 30 frames per second, and no audio.