lerobot/iamlab_cmu_pickup_insert

该数据集是一个机器人学数据集，包含631个片段，共146241帧，分为7个任务。每个片段包含一个视频文件和相应的信息文件。视频包含两种图像：全景图像和手腕图像。数据集还包括语言指令、机器人状态、动作、时间戳、片段索引、帧索引、奖励和完成状态等信息。数据集适用于机器人控制和视觉语言模型的研究。

This dataset is a robotics dataset containing 631 episodes, totaling 146,241 frames, divided into 7 tasks. Each episode includes a video file and an associated information file. The video contains two types of images: a full view image and a wrist image. The dataset also includes language instructions, robot state, actions, timestamp, episode index, frame index, reward, and done status. The dataset is suitable for research on robotics control and vision-language models.