hri_avoidance

该数据集是使用LeRobot工具创建的机器人技术数据集，专门针对so101机器人类型。数据集包含5个剧集，总计14744帧，覆盖1个任务，以30帧/秒的速率记录。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。数据集包括训练分割（剧集0到5）。特征包括：来自腕部摄像机的观察图像（480x640x3分辨率视频），机器人状态（6个浮点数，对应肩部旋转、肩部提升、肘部弯曲、腕部弯曲、腕部滚动和夹持器），手部位置（3个浮点数，对应x、y、z坐标），动作（6个浮点数，类似状态），任务阶段（1个浮点数，表示阶段），手部入侵（1个浮点数，表示活动状态），以及时间戳、帧索引、剧集索引、索引和任务索引等元数据。数据集适用于机器人学习、人机交互和避障等任务。This is a robotics dataset developed using the LeRobot toolkit, specifically tailored for the so101 robot platform. The dataset consists of 5 episodes, totaling 14744 frames, covering one task, and was recorded at 30 frames per second. The data is stored in Parquet file format, with a total data file size of 100 MB and a total video file size of 200 MB. The dataset includes a training split (episodes 0 to 5). The features contained in the dataset are as follows: observed images from the wrist camera (480×640×3 resolution video), robot states (6 floating-point numbers corresponding to shoulder rotation, shoulder lift, elbow flexion, wrist flexion, wrist roll, and gripper), hand positions (3 floating-point numbers corresponding to x, y, and z coordinates), actions (6 floating-point numbers analogous to the robot states), task phases (1 floating-point number indicating the current task phase), hand intrusion (1 floating-point number indicating the active status), as well as metadata including timestamps, frame indices, episode indices, sample indices, and task indices. This dataset is applicable to tasks such as robot learning, human-robot interaction, and obstacle avoidance.