|
In this paper, robots Motoman UP6 as foundation and build a industrial robots sorting system based on machine vision . The system consists of cameras , frame grabbers , computers, robotics and software . The system the job object image obtained by the camera , and analysis of images collected by the software , draw the coordinates of the target and classified information and maintain the target motion trail , and ultimately control the robot sorting operation . The research camera calibration technique , to achieve a direct linear algorithm in Matlab , and a world coordinate system to the robot coordinate system is an easy way to . The researchers then a variety of target detection algorithm focuses on target detection based on background subtraction and target detection based on the grayscale value of the two . In target recognition , linear discriminant function according to the characteristics of the job object area, perimeter , and training the classifier . Visual tracking , global nearest neighbor method with multi - hypothesis tracking method, programming the GNN algorithm and Kalman filter target tracking on the conveyor belt . Finally, on the basis of MOTOCOM32 research the Motoman robot PC-based control , so that the sorting program can control the robot to bring the workpiece conveyor continuous motion , real-time capture . In the course of the study also found that some shortcomings of MOTOCOM32 , and solutions are put forward . After several experiments and commissioning of completed research and construction of a primary industrial robots sorting system . Online experiments show that the results of this study , the system as a part of the industrial automation , has a good practical significance in the application , in theory also has some reference value .
|