| Welding has become an important metal process technology in modern industry because of its advantages of high reliability,high precision,low cost and so on.However,manual welding is difficult to meet the requirements of increasing welding speed and welding quality,and the arc light and smoke do great harm to the workers’ health who engaged in welding.Automation welding by robots is one of the solutions to solve the problems above,and it is also the trend in development of welding industry.Technology of welding seam detection is the key technology of automatic welding.The detection method based on structured light vision has the advantages of non-contact,high speed and high precision,and it has great value in research and application.In this paper,the key technologies of stainless steel fillet weld detection based on structured light vision are studied.The classification,development and research status at home and abroad of welding seam detection are studied.The camera model and the principle of optical triangulation are studied,and a structured light vision terminal based on CMOS industrial camera,structure light generator and filter is designed and built.The Eye-in-Hand assembly mode of terminal and robot is determined according to the requirements of the project.The calibration theory and method of the camera parameters,the position parameters of the structured light plane and the parameters of the position between the sensor and the robot are studied.The parameters are calibrated and optimized by the calibration toolbox of Matlab and the independent programs.Based on the analysis of the characteristics of the stainless steel fillet weld image,the image processing scheme composed of Region of Interest(ROI)selection,image filtering,feature detection and feature extraction is proposed,and the scheme is programmed by using Open CV library functions.The coordinates of the feature points in the robot base coordinate system are obtained through the coordinate transformation from the pixel coordinate system.The research contents of this paper are verified by experiments.The experimental results show that the maximum measurement error of the terminal based on structured light vision is less than ±0.145 mm,and the average error is 0.090 mm,The maximum deviation between the detected position and the reference position is less than 0.96 mm,and the average deviation is 0.35 mm. |
