面向轴承套圈的三维实时定位方法OA

To improve the positioning accuracy and matching efficiency of robotic arms in the grasping and stacking process of bearing rings,the research group proposed a stereo matching algorithm that integrates YOLOv8 with an optimized depth-based clustering method.Firstly,a binocular vision experimental platform was set up,and system parameters were calibrated using Zhang Zhengyou camera calibration method.Then,an improved stereo matching algorithm was applied to generate disparity maps and depth maps.Finally,a multi-environment bearing image dataset was constructed,combining object detection and depth estimation results.This enabled accurate bearing identification while converting depth information into point clouds,and the 3D coordinates of the bearings were calculated by perspective transformation matrices.Experimental results show that the proposed method can effectively improve positioning accuracy while maintaining real-time performance at frame rate of about 12 frames per second.The average absolute errors of the 3D coordinates along the x,y and z axes are 2.463,1.863 and 2.611 mm,respectively,and the errors meet the positioning accuracy requirements.