Research On Positioning Technology Of Valve Assembly Robot Based On Machine Vision

With the rise of industrial intelligence,more and more industrial robots appear in people's field of vision.The research on robots by major enterprises is also more in-depth and more comprehensive.Machine vision is an important part of robot research.Machine vision is one of the main ways to realize the precision positioning and recognition of industrial robots.This project mainly studies the vision positioning technology of the valve assembly robot.Relying on the existing laboratory equipment 2D-90 reciprocating compressor and ESTUN ER-30 six degree of freedom industrial robot,the recognition and positioning technology of compressor gas valve is studied to achieve the functional requirements of automatic assembly of gas valve.Firstly,the kinematics of the robot is analyzed to determine the relationship between the pose of the robot end and the joint motion parameters.The industrial camera used in the project is calibrated by using MATLAB software and OpenCV.Image processing algorithms such as image graying,image filtering and binarization are applied to realize the image processing of the compressor valve according to the special structure of the compressor valve,which lays the foundation for the subsequent assembly robot to identify and locate the compressor valve.Secondly,aiming at the problem of recognizing the gas valve edge contour by robot in the assembly process of 2D-90 reciprocating compressor valve,the edge detection method based on Canny operator is used to extract the gas valve edge contour information of the compressor,and the valve contour is successfully extracted,which realizes the visual guidance in the assembly process of the robot.The experimental results show that compared with the traditional methods,the proposed method can effectively suppress Gaussian noise,and the edge line is clear,without breakpoints,which can effectively improve the accuracy of edge location and meet the requirements of practical work.At the same time,Hough circle transformation and least square method are used to fit the contour of valve edge,which effectively improves the positioning accuracy.On this basis,based on the OpenCV machine vision library and visual studio 2017 software development environment,the valve visual positioning system is developed and designed by using MFC.The system has the functions of accurately identifying the position of the valve center,displaying the images collected and processed by the camera,as well as the relevant parameters of the camera.Finally,on the basis of the existing conditions in the laboratory,the vision positioning experimental platform of the valve assembly robot based on the above scheme is built by using the ESTUN ER-30 six degree of freedom Industrial Robot and 2D-90 reciprocating compressor.The recognition test and positioning accuracy test of the compressor valve are carried out by using the industrial robot vision system,and the valve assembly experiment based on the vision system is carried out successfully The accuracy and effectiveness of the vision positioning system are verified.