Research On Welding Trajectory Recognition And Control Technology Based On Structured Light Vision

The teaching and playback welding mode is bound to develop rapidly in the direction of intelligent autonomy with"high quality and high efficiency"as the core.The method of robot welding with the assistance of structured light vision is the most promising intelligent welding technology at present,and it can even be regarded as a bridge to upgrade the technology of"teaching welding-smart welding-autonomous welding".In the current welding field,welding trajectory recognition and control technologies of laser multi-point positioning,pre-welding trajectory fitting,and welding seam real-time tracking,are the three implementation forms of structural light vision-assisted welding.To this end,in the context of the era of intelligence,this paper discusses the above three welding modes for key technologies such as image processing,trajectory recognition,and trajectory control.In this paper,the software and hardware system of the teaching and playback welding is firstly designed and developed,which provides a hardware operation platform and software development platform for the subsequent research of three welding trajectory recognition and control technologies.The designed teaching welding control software can control the welding platform to complete a series of the teaching and playback welding operations.The system of the teaching and playback welding has functions such as manual teaching and automatic welding.The weld image processing is the core problem of three welding modes:laser multi-point positioning,pre-welding trajectory fitting,and real-time welding seam tracking.Based on the current research theory of image preprocessing,various methods are measured and explored,and the image preprocessing process with median filtering,the Ostu threshold segmentation,morphological processing of closing and opening operations,and laser stripe bone line extraction by Gray weighted centroid algorithm as the core is determined.at the same time,various laser stripe bone line models that may appear under actual working conditions are analyzed,and a straight line fitting intersection theory that can adapt to the actual conditions and accurately obtain weld feature points is proposed.Removing abnormal values of characteristic points of welding seam is an important method to ensure welding accuracy and safety.Reflection,refraction,mirror impurities and other factors will inevitably cause some abnormal points to appear irregularly.In this context,this paper proposes a segmental outlier removal theory,which is embedded in the welding mode of pre-welding trajectory fitting and real-time welding seam tracking.It ensures the smooth and reliable trajectory operation,further improves welding accuracy.This paper discusses in detail the technical characteristics of three welding modes:the laser multi-point seeking welding mode,the pre-welding trajectory fitting welding mode,the welding seam real-time tracking welding mode,and proposes trajectory identification and control methods corresponding to the above three welding modes:correcting the teaching trajectory by laser positioning,quasi-uniform cubic B-spline curve fitting welding trajectory,and real-time correction of welding seam tracking.Finally,based on the teaching welding control software,according to the image processing method,the welding seam feature point outlier removal theory,and the trajectory recognition and control method,three sets of welding trajectory recognition and control software:the laser multi-point seeking welding mode software,the pre-welding trajectory fitting welding mode software,the welding seam real-time tracking welding mode software,are designed and developed.Their functions were tested on the welding platform.The actual measurement proves:when welding in laser multi-point positioning and pre-welding trajectory fitting mode,the actual welding trajectory is basically consistent with the centerline of the weld,and the trajectory recognition accuracy is on the order of 10^-2mm;when welding in the seam real-time tracking mode,the tracking real-time deviation averages are controlled within the range of±0.3mm;the three sets of welding track identification and control software developed basically meet the requirements of actual welding production.