Research On Visual Tracking System Of Right Angle Welding Seam

There are a large number of right-angle welds in major lifting equipment,shipbuilders,and steel structure workshops.However,due to the inflection point in the right-angle weld and the poor accessibility of the welding torch,the processing is difficult.At present,manual welding is still used to achieve right-angle welding.Seam welding,welding quality and efficiency need to be improved.Therefore,this paper studies a real-time tracking system based on visual sensing,which realizes the automatic tracking welding of rightangle welds,thereby improving the quality and efficiency of right-angle welding.The main research contents of this paper are as follows.This paper firstly builds a visual tracking platform for right-angle welds,mainly including arc welding robot system and visual sensing system.By comparing the performance parameters and adaptability of each hardware,the automatic welding equipment,laser vision sensor and other components The model was selected,and the hand-eye calibration was carried out using the two-dimensional plane target calibration method.Based on the Socket Message protocol,a real-time data communication scheme between the robot,the vision sensor and the PC is designed.This paper designs a set of welding seam image processing methods,including image filtering,laser stripe centerline extraction,welding seam feature point extraction and other modules.Aiming at the extraction of laser stripe centerline,the current laser stripe centerline extraction technology is analyzed,and an optimized gray-scale centroid method is proposed;for the extraction of weld feature points,a weld feature point based on SUSAN operator is used The extraction method has good robustness and uses fewer computational parameters,and the computational efficiency is significantly improved.This paper studies the welding seam tracking algorithm and the path planning method at the inflection point.Aiming at ABB IRB1410 robot,combined with its D-H parameters,the kinematics modeling was carried out,and its forward and inverse kinematics were deeply studied to realize the robot motion control of right-angle welding seam visual tracking system.The deviation is predicted,and the target state equation and observation equation of the system are established at the same time;the path planning is carried out at the corner point of the right-angle weld,the right-angle weld is divided into a straight welding area and a welding torch attitude adjustment area,and the welding torch is adjusted by estimating the normal vector of the corner point.attitude.In order to verify the feasibility of the right-angle welding seam visual sensing system,several sets of comparative experiments are designed.The experimental results show that the system can reduce the interference effect of noise on the welding seam tracking,improve the stability and accuracy of the welding seam tracking,and analyze the experimental results.According to the data,the distance from the detection point of the tracking sampling to the ideal trajectory is less than 2mm;the welding seam tracking experiments were carried out on the chamfered welds with three chamfered angles of 90°,110°,and 120° respectively.The tracking method has good adaptability.