Research On Welding Technology Based On Vision Guidance Of Line Laser Camera

The use of welding robots for automated welding in industrial applications is of great significance.The welding robots in the existing process production can basically achieve straight or curved welding on the specified plane,but the goal of welding trajectories on different planes is still unable to get a better solution.In response to the above problems,the paper puts forward a scheme of automatic welding system based on line laser camera by introducing machine vision technology.The system first uses the line laser and CCD(Charge-coupled Device)industrial camera to restore the three-dimensional shape of the workpiece to be welded,and then realizes the automatic welding of the workpiece weld and the evaluation of welding quality through image processing and three-dimensional path planning.The welding path planning and model reconstruction methods described in the paper have a certain promotion effect on the improvement of automated welding technology.The main research contents of the paper are as follows:(1)Introduce the research background of current automatic welding technology,analyze the shortcomings and disadvantages of manual welding,introduce the concept of automatic welding,and then introduce the development trend and future development direction of automatic welding robots.Finally,the status quo of the current 3D point cloud technology is analyzed,and the description of the 3D information of the object is realized by replacing the 3D camera with a line laser and a 2D camera.(2)Analyze the principle of the line laser camera system and establish a camera model based on perspective transformation.Then it introduces the method of extracting the center line fringe of the line laser,and image processing the laser fringe image,including the introduction of mean,median,Gaussian,bilateral filters and comparison of filtering effects.Finally,the traditional centerline fringe extraction method is analyzed and a new improved centerline extraction method based on Steger method is proposed.The processing of the centerline of the laser fringe is completed,which lays the foundation for the accuracy of the point cloud below.(3)Study the use of Zhang Zhengyou’s camera calibration method to perform camera calibration experiments,which solves the problems of accuracy and efficiency of traditional calibration methods and improves the efficiency of camera calibration.Based on Zhang Zhengyou’s checkerboard calibration,the calibration of the line laser plane is integrated.A checkerboard image is used to match a line laser image to solve the points on the laser plane in the camera coordinate system.Finally,the least squares are used to fit the plane.Method to solve the laser plane equation.Then the robot tool coordinate system calibration and the camera and robot hand-eye calibration experiment were realized.(4)The method and process of point cloud preprocessing are studied.Including points cloud average distance calculation,neighborhood solving,normal vector estimation,and redundant point filtering.Then the depth map projection is performed on the point cloud and the ground,the contour extraction based on the Canny operator is performed on the workpiece in the two-dimensional image,and the inner and outer contours are obtained.The coordinates of the intersection point.Finally,the three-dimensional coordinates of the intersection point are obtained according to the projection matrix inverse to complete the path planning of the welding point.(5)Designed the overall structure of the automatic welding system based on the line laser camera,and introduced three parts:the vision part,the robot execution part,and the hardware selection and design of the welding part.Then introduced the software design and operation management of this project.Finally,the success rate of welding test on passenger car pedals can reach 87.5%,and the electromagnetic induction method is used to realize the analysis of welding quality.