Research On Weld Inspection And Control System Based On Binocular Vision Manipulator

With the continuous improvement of welding quality and efficiency requirements in the industry,robot welding technology has been more and more widely applied in the manufacturing industry.In this context,the realization of intelligent welding process will be an inevitable trend of robot welding technology.However,at present,the most widely used teaching robot has many problems such as cumbersome operation,low efficiency and precision depending on the experience of the instructor.Furthermore,the off-line programming robot carries out virtual simulation,which lacks the perception of the real environment and often leads to non-ideal welding precision.In response to this,in this paper,binocular vision technology is applied to the field of welding joint detection by means of mechanical arm,thus achieving the purpose of high quality,high efficiency and high flexibility.Firstly,the forming model and distortion model of the camera are analyzed,the relationship between coordinate systems is established,the calibration method is analyzed systematically,and the camera is calibrated,thereby obtaining the internal and external parameter matrix of the binocular camera and establishing the relationship between the binocular camera and the world coordinate system.Secondly,the images acquired by the camera are processed,two algorithms for detecting fillet welding joints are developed,and its threshold segmentation,filtering,edge detection,morphological processing and transformation,rectangle fitting and the like are processed,thus recognizing the position of welding joints accurately in complex environment.Next,the matching experiment is carried out,and the feasibility analysis is carried out on the matching results.Then,the 3D coordinate formula of the objects under test is deduced by the camera model under the parallel optical axis,and the least square method is used to calculate the value of depth direction more accurately.On the basis of this ideal model,the condition of non-parallel optical axis is further studied,and the workpiece coordinates in the world coordinate system are converted into the coordinates in the mechanical arm’s own coordinate system.Finally,the recognition system of fillet welding joints is constructed,the hardware selection is carried out according to the goal to be achieved,the software system based on Python and OpenCV is designed,the feasibility of the system is verified by experiments,and the causes of errors are analyzed.This paper develops a set of weld inspection system based on Python 2.7 and OpenCV3.0,which can realize the function of detecting the position of the fillet weld,and through the results of the welding simulation experiment,the system basically meets the expected requirements,this system meets the application requirements.