Research On Key Technology Of Weld Location Detection Based On Laster Vision System

Automatic welding technology based on industrial robots is one of the important research directions in industrial automation.At present,most welding robots use the traditional teaching and reappearing method to obtain welding seam information,which is greatly influenced by human factors.When facing the welding seam with complex trajectory,it is easy to find that the actual welding track has a big deviation from the preset teaching route.Weld seam location detection technology can solve this problem effectively.Weld detection technology based on laser vision system has important research value because of its high precision,high efficiency,non-contact and other advantages.Based on the detailed analysis of seam detection schemes at home and abroad,a seam detection system based on laser vision sensor was designed.(1)Laser vision sensor was the core part of the system.After the analysis and research,the laser vision sensor based on laser source,industrial camera,lens,filter and other devices was built,the appropriate optical structure was selected and the software platform was designed to obtain clear and stable weld image information in complex environment.(2)Aiming at the laser seam image acquired by sensors,the image processing algorithm was studied.The mathematical model of extracting laser fringe skeleton was abstracted by a series of pretreatment methods.On this basis,genetic algorithm was used to improve the efficiency.Finally,after eliminating noise points iteratively,the extracted data was fitted by a straight line to obtain accurate position data of laser fringe skeleton and the coordinates of weld feature points.(3)The application of related calibration theory and method.Using camera calibration,laser plane calibration and robot hand-eye calibration,the transformation relationship of weld feature points from two-dimensional image system to threedimensional space coordinate system was obtained.(4)The function test experiment and precision evaluation experiment of the weld seam detection system based on laser vision sensor were carried out.The image processing module and system calibration module were tested separately,and the calibrated system was used to detect the known overlap welds with straight lines and curves,and the overall average deviation of the system is less than 0.50 mm,and the average detection time was 0.46 s.The experimental results showed that the system can overcome the interference of welding environment,identify the weld seam accurately,and meet the requirements of practical application.