ZhaoRunyi/Piper_open_pan_0421

Piper LeRobot数据集是一个机器人操作数据集，使用ALOHA双机械臂系统采集。数据集包含左右机械臂的关节状态（各6个关节）、夹爪状态、末端执行器位置（3维）和旋转（rot6d格式，6维），以及来自三个摄像头的图像数据（高清摄像头、左腕摄像头、右腕摄像头）。数据以10 FPS采样，采用完整采样模式，并以高清摄像头的时间线为基准进行对齐。数据集通过转换Piper HDF5格式生成，适用于机器人模仿学习、强化学习等任务。

The Piper LeRobot dataset is a robotic manipulation dataset collected using an ALOHA dual-arm robot system. It includes joint states (6 joints per arm), gripper states, end-effector positions (3D), rotations (in rot6d format, 6D), and image data from three cameras (high camera, left wrist camera, right wrist camera). Data is sampled at 10 FPS in full sampling mode, aligned based on the high camera timeline. The dataset is generated by converting Piper HDF5 data and is suitable for tasks such as robot imitation learning and reinforcement learning.