Development Of Autonomous Programming System For Welding Robot Based On Stereo Vision

Under the background of intelligent manufacturing,the traditional manufacturing industry urgently needs to introduce advanced technology to achieve industrial upgrading and improve the degree of automation and intelligence.In the field of welding robots,the realization of intelligent welding through vision guidance is becoming a current research hotspot.This thesis takes the integration of the autonomous programming system of welding robots based on binocular vision as the main research goal.Systematic development was carried out around platform construction,welding seam trajectory identification,welding seam 3D trajectory extraction,hand-eye calibration,real-time communication,coordinate transformation,and self-programming system series and integration.It mainly solves the following four problems.The idea of welding seam trajectory extraction based on two-dimensional image was proposed,which can accurately extract the pixel coordinates of the twodimensional trajectory image of the butt straight weld.Through the methods of image acquisition,image filtering,threshold segmentation,morphological processing,edge detection,workpiece contour recognition,welding seam centerline identification and welding seam trajectory extraction,the three straight lines connecting the short side of the workpiece contour and the welding seam centerline were intersected.The method of extracting the starting point and end point of the welding seam,and discretizing the starting point and the end point,and extracted the coordinate value of the welding seam trajectory in the image pixel coordinate system.And verify the stability of the weld extraction method.A three-dimensional reconstruction method of welding area based on SGBM(semi-global block matching)stereo matching was proposed,and the threedimensional information of the discrete trajectory of the weld in the camera coordinate system was obtained by calculation.Using the MATLAB binocular calibration toolbox based on Zhang’s calibration method,the internal and external parameters of the binocular camera were calculated.By comparing the disparity map effects of BM(Block Matching)and SGBM stereo matching,the SGBM stereo matching algorithm was selected as the basis for subsequent development;the holes generated in the disparity map were filled by means of mean filtering and restored to a3 D point cloud display.Input the two-dimensional trajectory coordinates into the SGBM stereo matching disparity map,extract the three-dimensional information of the weld trajectory in the camera coordinate system,and verify the error of the binocular vision system by comparing the calculated and measured values of the weld length.A series method of autonomous programming system was proposed.Using eyeon-hand hand-eye configuration,the hand-eye calibration results based on the Tsai two-step method were calculated by MATLAB,and the hand-eye calibration error was quantified by the reprojection method;the UDP real-time communication between the computer and the control cabinet was constructed to obtain the robot pose in real time.The transfer matrix was constructed for the three-dimensional information of the robot pose,hand-eye calibration result,and welding seam trajectory,and the welding seam position information was successfully transformed into the robot base coordinate system,and the welding attitude of the robot perpendicular to the workpiece was calculated.Quantitative analysis of the errors existing in this system.Finally,the autonomous programming system software was modularized,the user interface was built,the key indicators of the system were quantified,and the stability of the system was tested.The positioning error of the vision-guided robot was calculated by comparing the values of the feature points of the positioning welds of the vision-guided robot with the values of the feature points of the positioning welds manually taught.The experimental results show that in the robot base coordinate system,the positioning error of X-axis and Y-axis was less than 1mm,and the positioning error of Z-axis was less than 5mm,and the error value meets the initial positioning requirements of actual welding.