black-foam-redX_2-80eps_20260513_114247

该数据集是使用LeRobot创建的机器人数据集，专门用于机器人学习和控制任务。数据集包含80个episodes，总共40700帧，以30fps的帧率录制。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。数据集的特征包括：机器人的动作（action），由6个浮点数值表示，对应肩部平移、肩部提升、肘部弯曲、手腕弯曲、手腕旋转和夹爪位置；观测（observation）包括状态（state），同样由6个浮点数值表示机器人的关节位置，以及三个视角的图像：顶部视角（top）、手腕视角（wrist）和D435i相机视角（d435i），每个图像为视频格式，分辨率480x640，3通道，使用AV1编解码器。此外，数据集还包含时间戳、帧索引、episode索引、索引和任务索引等元数据。机器人类型为so_follower，适用于训练和评估机器人控制模型。数据集仅包含训练分割（train），覆盖所有80个episodes。This dataset is a robotic dataset created with LeRobot, specifically designed for robotic learning and control tasks. It contains 80 episodes, totaling 40700 frames recorded at 30 fps. The data is stored in Parquet format, with the total size of the Parquet data files being 100 MB and the video files being 200 MB. The dataset features include: the robot's action, represented by 6 floating-point values, corresponding to shoulder translation, shoulder lift, elbow flexion, wrist flexion, wrist rotation, and gripper position; the observation includes the state, which is also composed of 6 floating-point values representing the robot's joint positions, as well as images from three perspectives: top-view (top), wrist-view (wrist), and D435i camera view (d435i). Each of these image streams is in video format, with a resolution of 480×640, 3 channels, and encoded using the AV1 codec. In addition, the dataset also includes metadata such as timestamps, frame index, episode index, index, and task index. The robot platform for this dataset is so_follower, which is suitable for training and evaluating robotic control models. The dataset only includes the training split, covering all 80 episodes.