DORLR/ur5_real_scripted_rl_merged

该数据集由LeRobot创建，是一个用于机器人技术的数据集，专门针对UR5机器人。数据集包含641个训练集片段，总计141637帧数据，覆盖3个不同任务。数据以30fps的帧率采集，包括两种摄像头视角的图像数据：手腕摄像头（分辨率为480x640，3通道）和侧面摄像头（分辨率为720x1280，3通道）。此外，数据集还提供机器人的状态观测（如关节位置）和动作数据（如关节控制命令），这些数据以浮点数组形式存储，涉及7个关节（包括机器人夹爪关节）。数据集结构以Parquet文件格式组织，适用于机器人学习任务，如强化学习或模仿学习。This dataset was created using LeRobot and is designed for robotics applications, specifically for the UR5 robot. It consists of 641 training episodes, totaling 141637 frames, and covers 3 distinct tasks. Data is collected at 30fps and includes image data from two camera perspectives: a wrist camera (resolution 480x640, 3 channels) and a side camera (resolution 720x1280, 3 channels). Additionally, the dataset provides robot state observations (e.g., joint positions) and action data (e.g., joint control commands), stored as float arrays for 7 joints (including the robot gripper joint). The dataset is organized in Parquet file format and is suitable for robot learning tasks such as reinforcement learning or imitation learning.