Research On Weld Recognition And Trajectory Planning Based On Machine Vision

Along with the continuous progress of "made in China 2025",manufacturing industry develops rapidly in the direction of intellectualization.Welding technology occupies a very important position in modern manufacturing industry,and the intelligent welding process is the future development direction of welding technology.At present,most welding robots uses manual teaching to carry out welding operations,which are inefficient,weak adaptive ability,and can not ensure the welding quality.Aiming at this problem,this paper researches weld seam recognition and seam tracking based on machine vision.The detailed research work are as follows:(1)Overall framework design of the system.Combined with the traits and design requirements of the system,Analyzed and studied on operating principle of the camera,lens and light source,and selected based on.Made brief introduction of the main body,control device,teaching device for selected ABB IBR1410 robot.Then the relationship between welding process parameters and weld quality is discussed.The Visual Studio2019 and Open CV4.1.0 image processing environment is configured finally,and imported seam tracking analysis for the robotic toolbox of Robot toolbox.(2)Calibration of vision system.The model of camera imaging process is deduced,and the method of camera calibration is introduced,the method of a flexible new technique for camera calibration is selected and solving principle of this method is deduced;the internal parameters and distortion coefficient of the camera are obtained through camera calibration experiment,and calculated overall average re projection error is 0.19 pixel in accordance with the calibration result;and obtained eye-in-hand matrix by its calibration.(3)Test of weld center line.Firstly,the pre-process the colorized weld image to gray,enhancement,denoising,aiming on weld seam image of this paper,that obtained grey-stretching algorithm and median filtering algorithm have better enhancement,denoising effects through results of different methods.Then,aiming at the problem of welding seam and background segmentation,this paper proposed the processing method by combination of binarization and closed operation of morphologic operation,which aims at segmentation of weld and background,which segment the weld seam from the background area.Finally proposed grey weighted centroid method line-by-line of adaptive weld width,and found continuous weld center line.(4)Research on the moving tracking of welding robot.Built D-H parameter model of ABB IBR1410 welding robot,and obtained its forward kinematics,inverse kinematics equations and workspace;in order to ensure the continuity of robot welding process,under Cartesian workspace,proposed planning method for the sharp angle of parabola over inflection point is proposed for the broken line weld,and the method of cubic spline planning is adopted for curved weld,that the average tracking tolerance for both two planning methods are 0.16 mm and 0.14 mm,which meets the requirement of weld;and motion simulation results of robots shows both two methods can ensure the continuity of the robot's moving tracking in space.