yinongh/real_world_automate_1
收藏资源简介:
这是一个由LeRobot创建的机器人操作数据集,专注于插头插入任务。数据集包含16个训练片段,总计1097帧数据,帧率为30fps。数据采用多模态格式,包括机器人末端执行器动作(9维浮点数,含位置和旋转变化)、观测数据(如末端执行器内部力、姿态矩阵、对齐的插座深度图像、初始插头深度图像、点云数据)以及来自多个摄像头(手腕摄像头、Azure Kinect前摄像头、辅助摄像头)的RGB和深度视频流。数据集结构包含时间戳、帧索引、任务索引等元信息,适用于机器人模仿学习或强化学习研究。
This is a robotics manipulation dataset created using LeRobot, focusing on a plug insertion task. It contains 16 training episodes with a total of 1097 frames at 30fps. The dataset features multimodal formats, including robot end-effector actions (9-dimensional float values for position and rotation changes), observations (such as end-effector internal forces, pose matrices, aligned socket depth images, initial plug depth images, point cloud data), and RGB/depth video streams from multiple cameras (wrist camera, Azure Kinect front camera, auxiliary camera). The dataset structure includes metadata like timestamps, frame indices, and task indices, making it suitable for robot imitation learning or reinforcement learning research.




