| Optical vision sensor system based on laser marking is a commonly used welding seam tracking system.In common,one-line laser line is used to identify welding seam offset.In the existing 2D welding seam tracking system,its algorithm is simple and fast,but it relies heavily on the relative position of laser line,camera and welding gun,and is not adaptable to all kinds of welding seams.In common 3D vision recognition,a large number of advanced image processing algorithms are used.Although they have strong adaptability to all kinds of welds,the computing system for image processing has high performance requirements and high price,and cannot be integrated into the microcontroller chip of conventional industrial control system.On the basis of studying the basic application and algorithm of 2D and 3D welding seam laser tracking,this paper puts forward a method to improve the existing 2D and 3D welding seam tracking based on reticle laser,designs a simple platform structure,and compiles the graphics processing algorithm and application based on STM32 microcontroller.The reliability and stability of the algorithm running on STM32F107 microcontroller are studied by cross experiment.The research results show that the processing of images with pixels of320*240 on STM32F107 microcontroller can identify the deviation between weld and laser center point,and the execution period of the algorithm is less than 0.15 s s.Integrated with the original welding table motion control program,the online feedback control system controlled by a single chip can be realized,and the overall execution cycle can be controlled within 0.2s;The feasibility,reliability and stability of welding seam deviation identification and stepping motor feedback control,through three sampling frequencies of camera 1Hz,2Hz and 4Hz,the welding gun position adjustment speed is constant speed and variable speed mode,and the cross experiment shows that the system has the greatest fluctuation in 1Hz constant speed mode,and both 2Hz and 4Hz variable speed modes have good image recognition rate and fast welding gun position adjustment ability.Under the control of simple cross slide table and ordinary stepping motor,the steady-state error of the system is less than 0.2mm,which can For the reflection problem of weldment,the incident angle of laser can be adjusted to make it deviate from the effective position of camera processing image,and the weld recognition is basically undisturbed after processing by this method;The program identification needs to be further improved and verified when simulating the strong light in the welding state,but by adjusting the shielding position of the shielding plate,the effect can be basically unaffected.Machine vision welding seam tracking method based on reticle laser can effectively improve the defect that the original 2D tracking system relies heavily on the relative positions of laser line,camera and welding gun,and at the same time greatly simplify the 3D welding seam tracking image algorithm,so that it can run on STM32 and other conventional controllers,thus reducing the overall cost of welding seam tracking sensor and control system,making it easier to combine sampling and control closely,and having great practical value. |