Research On Laser Vision 3D Guidance System For Pipeline Welding Robot

Welding is one of the most important processing methods in modern mechanical manufacturing industry.It has become an inevitable trend to use intelligent robot to realize automatic welding.At present,the research of pipeline welding robot seam tracking is mainly based on 2D seam images.It is difficult to obtain important parameters such as seam depth and groove angle,and it is difficult to achieve high-precision seam tracking for pipelines of different diameters.Based on the principle of structured light vision measurement,a laser vision 3D guidance system for pipeline welding robot is designed in this paper.The system can accurately obtain the 3D information of the weld and guide the torch to track the weld in real time.The main contents of this paper are:1.A laser vision 3D guidance system for pipeline welding robot is designed,which consists of a laser vision sensing system and a motion control system.Laser vision sensor system is mainly used for collecting and processing weld image and calculating 3D information of weld;motion control system controls welding crawler creep,welding torch swing and welding torch height adjustment to realize 3D motion control of welding torch.Modbus protocol is used to communicate between laser vision sensor system and motion control system to realize real-time seam tracking.2.According to the characteristics of pipeline welding robot,the principle of laser vision 3D measurement for pipeline welding robot is studied.The mathematical model of 3D shape measurement of weld seam is established,and the calibration method of camera internal parameters and light plane equation parameters of the model is given.3.Aiming at the seam image strongly disturbed by welding arc light and welding slag spatter,the seam stripe image processing algorithm is studied.A directional template threshold segmentation algorithm is proposed.Different templates are used according to the direction of the light strip in the weld image to calculate the results.The multi-level threshold is used to judge the results,and a better image segmentation result is obtained.The gray scale centroid method and slope analysis method are used to extract the weld feature points accurately from the image segmentation results.It takes less than 0.1s to process an image and calculate the 3D information of weld seam,which meets the real-time requirement of seam tracking.The processing results of multiple weld images show that the correct extraction rate of weld feature points can reach 99%.4.Weld tracking experiments were carried out on various welds of pipelines in factory environment.The results show that the weld surface of pipelines is well formed,the lines are uniform,and the weld tracking is accurate.