Research On Weld Seam Recognition And Tracking Technology Of Welding Robot

Welding robot is of great value to the automation of welding process.The existing welding robots can realize automatic welding of straight line welds,but there are still some difficulties for curved welds.In order to solve the problems above,this paper introduces machine vision technology and proposes a welding robot system scheme which can recognize and track the curved weld seam.Firstly,image processing algorithm is used to identify the coordinates of feature pixels,and then visual guidance is used to realize automatic welding and welding quality evaluation of curved welds.The work of this paper is of positive significance to improve the level of welding equipment in Chinese manufacturing industry.The main research contents of this paper are as follows:(1)System scheme design: According to the actual needs of vision-guided welding robot,the overall scheme of the system is determined.This paper elaborates the vision guidance and non-destructive testing scheme,and chooses the relevant hardware equipment.Visual guidance adopts monocular vision to detect weld seam,and electromagnetic sensors are used to evaluate welding quality for non-destructive testing.(2)Image processing: image processing is carried out on the images collected during welding process.Median filtering is used to smooth the images.Deriche operator is used to extract image edge features.Then high-speed area filtering algorithm is used to further remove the interference.Finally,histogram segmentation and gamma correction are used to enhance the image.(3)Weld path planning: In the pixel coordinate system,the image pixels are identified and processed by Opencv,and the coordinates of the upper and lower lines of the weld are obtained,and the coordinates of the central line of the weld are further extracted.Hermite interpolation algorithm is used to obtain the coordinates of the initial points.Welding position and posture are 75 degree on the plane of torch and weld seam and 90 degree on the normal of torch and weld joint.The transformation angle of welding robot joint is obtained by adjusting coordinate system and expecting coordinate system transformation algorithm,and the welding coordinates to ensure welding posture are obtained.Through camera calibration,tool coordinate system calibration and hand-eye calibration,the mapping from two-dimensional pixel feature points to three-dimensional entity coordinates is completed.(4)Software design: Develop the application software of welding robot seam recognition and tracking system,including hardware communication interface,calibration interface,manual welding interface,automatic welding interface and welding evaluation interface.The welding evaluation interface presents the welding quality inspection processmore intuitively in the way of real-time data curve,and the software gives the evaluation results for the operator to analyze and process.(5)Systematic experiment: Through welding experiments on irregular curves and straight steel plates,the deviation average error between the center line of curved weld and the center line of weld pool is 0.6 mm,and the curve of weld seam is smooth,which meets the requirements of welding error.The experiment proves that the system has high precision response ability to irregular weld and straight weld,and has strong applicability.