Research On Thin Plate Butt Welding Seam Tracking System Based On Vision Sensing

Sheet metal sheets are widely used in marine,machining,aerospace and other industries.At present,the welding quality is unstable due to manual operation,and the technical level of welders is uneven,making automatic and intelligent welding the mainstream trend of future development.Due to the light reflection,splash,dust and other noises generated during the automatic welding of thin plates,the position information of the welding seam will be blocked,thus affecting the identification and extraction of feature points.Therefore,for the I-type welding seam of thin plate butt joint,the method of welding seam feature recognition based on visual sensing technology is studied,which can accurately extract the welding seam feature points,which lays a foundation for the follow-up tracking work.The main contents of the paper are as follows:(1)Establish a set of welding seam feature recognition system based on visual sensing.The structure and workflow of the vision system are understood in detail,the composition and selection of the vision sensor,and the position design of the sensor are studied.Calibration and analysis of visual sensing system.Learn the methods,steps and principles of camera calibration and hand-eye calibration of robots,camera calibration and hand-eye calibration experiments were performed.Combined with the mathematical model of the laser vision sensor and the imaging principle of the camera,the mapping relationship between the image coordinates of the weld fringe and its corresponding world coordinates is established.(2)In this content,a set of effective welding seam image processing method is studied for the I-type weld of thin plate butt joint.High-efficiency weld image processing algorithms including extracting regions of interest from images,image denoising,image segmentation,and identification and extraction of feature points.In order to solve the problem of more noise in the welding process,a connected area algorithm is proposed to identify and mark the weld feature points,and the connected area algorithm is improved to extract the weld feature points,and the identification and extraction of the weld feature points are successfully realized.(3)Write a welding seam feature recognition system.On Windows10 system,combined with Pycharm2017 and Qt development platform,based on Opencv visual open source library,a special software for welding visual image processing and coordinate transformation was developed.Its functions include image filtering,image binarization,feature point recognition and extraction,and feature point coordinate conversion.By comparing the gap width of the thin plates,it is detected that the average error of the welding seam tracking system based on visual sensing is within 0.067 mm,which improves the reliability and practicability of welding seam feature extraction and meets the real-time requirements of welding.It also shows that the proposed algorithm can reduce the influence of noise on the visual sensing system,improve the accuracy and stability of the system,and is suitable for thin plate docking,and the whole system meets the needs of actual robot welding production.