Model Information Based Structured-light Vision Correction System For Arc Welding Robot

At present,automation and intellectualization is the trend of welding production.Various kinds of automatic equipment are introduced to the welding process,among which the most widely used equipment is arc welding robot.However,most of the arc welding robots work in "teach-playback" mode which is not suitable for welding of non-standard parts.To improve the flexibility of arc welding robot,a model information based structured-light vision correction system for arc welding robot is studied.According to the visual simulation information and model trajectory information obtained from the robot offline simulation program,the method of weld positioning and trajectory correction are proposed.As a result,the robot can adapt to the welding of non-standard parts.In order to obtain accurate prior model information,the process of parameters synchronization between simulation and actual robot workstation is firstly executed.On this basis,the simulated laser stripes which can reflect the shape and position of the actual laser stripes are obtained from the simulation environment as the prior information of image processing.Then,two kinds of image processing methods are proposed: simulation image template matching method,simulation noise reduction registration method.The first method uses the simulation laser stripe to generate the image template,and then uses the shape template matching method with high robustness to position the weld seam.After investigating the influence of different parameters on the recognition effect,the performance of the proposed method is evaluated and analyzed considering the recognition time,recognition accuracy and anti-interference,and compared with the previous methods.The second method divides region of interest(ROI)according to the shape and position change rule of simulation laser line to filter out most of the interference,and then constructs structural elements according to the normal direction of simulation laser line and makes further noise reduction by non-uniform morphological corrosion.After skeletonizing the laser stripe,the seam can be positioned by registering the simulation laser stripe to the skeleton.Then,the effect is similarly evaluated considering the recognition time,accuracy and anti-interference,and compared with the previous methods.Based on the initial model trajectory provided by the offline simulation program,the segment and correction strategy is designed for the typical non-standard parts with discontinuous multi-section welds,and then experiment is executed.In order to improve the production efficiency and adapt to the thermal deformation in the welding process,an online trajectory correction strategy based on PID controller is designed,and an appropriate data storage strategy is introduced to adapt to the working characteristics of structured light scanning.The typical weld seams in shape of line and circle are selected for welding experiment,and good track correction effects are obtained.In order to avoid welding deviation caused by recognition deviation points in real-time seam tracking,a strategy based on trajectory estimation is proposed to eliminate the recognition deviation points.Experimental results show that this strategy can effectively filter out the recognition deviation points,so that the welding deviation will not occur.