Aasdfip/hri_1

该数据集是一个机器人控制数据集，由LeRobot创建，专门用于双臂机器人（bi_so_follower类型）的模仿学习或强化学习任务。数据集包含13个episodes、11803个frames和3个不同任务，数据以parquet文件格式存储，总数据大小为100MB，视频文件大小为200MB，帧率为30fps。数据集的特征包括：动作（action）和观察状态（observation.state），均为12维浮点数组，对应左右臂的肩部、肘部、腕部和夹爪的关节位置；观察图像（observation.images）来自三个摄像头视角（左腕、右腕和右上方），每个视频的分辨率为480x640，RGB三通道，使用AV1编码；此外，还包含时间戳、帧索引、episode索引、索引和任务索引等元数据。数据集仅包含训练集，适用于机器人控制算法的训练和评估。

This dataset is a robotics control dataset created using LeRobot, specifically designed for imitation learning or reinforcement learning tasks involving a dual-arm robot (type: bi_so_follower). It includes 13 episodes, 11803 frames, and 3 distinct tasks, with data stored in parquet files totaling 100MB and video files totaling 200MB at a frame rate of 30fps. The dataset features: actions and observation states as 12-dimensional float arrays representing joint positions for left and right arms (shoulder, elbow, wrist, and gripper); observation images from three camera perspectives (left_wrist, right_wrist, and right_top), each video with a resolution of 480x640, RGB channels, encoded in AV1; additionally, it includes metadata such as timestamp, frame index, episode index, index, and task index. The dataset only contains a training split and is suitable for training and evaluating robot control algorithms.