Research On The Weld Seam Tracking Trajectory Control Technology Based On Structured Light Vision

From the traditional manual welding to automatic,programmed,intelligent implementation of welding is an inevitable trend in the development of welding.The current industrial welding robots mostly use the "teaching-reproduction" welding pattern.This mode still requires manual instruction.To make welding intelligent,it is necessary to study the self-recognition of trajectory welding mode without manual instruction.In this paper,we present a structured light-based tracing and tracking control mode for the weld seam,and study the key technologies involved in detail.The hardware and software platform is first built to implement the tracer tracking welding.The functions of each component unit and its role in the overall system architecture are analyzed.On this basis,the design and construction of the hardware platform,including the structured light vision sensor,is completed,and the functions and algorithms of each module in the software system are developed to form a complete system of integrated algorithms that can realize the trace tracking control of welding.Then,a model is built to transform the pixel coordinate of the pixels on the welding seam image to the three-dimensional coordinates in the actual space under the robot base coordinate system,in such a way that the pixels on any line of structured light in the welding seam image can be transformed to the spatial coordinates.This model is used to translate the pixel points on any line of structured light into spatial coordinate.Internal and external parameters from the camera calibration are required,followed by calculations for the structural optical surface calibration and the hand-eye calibration to integrate the conversion model and verify its accuracy.The image treatment of the weld seam is essential for the whole tracking trajectory welding.On the basis of the existing image processing flow of the structured light weld seam,we analyze the effect of each algorithm from smoothing,threshold segmentation,region of interest extraction,morphological processing,connected region extraction,structural light stripe centerline extraction,to weld seam shape feature point extraction on the structural light weld seam image processing,and obtain the optimal image processing system.Finally,the tracking trajectory control system algorithm of the trajectory pattern recognition research,on the basis of system calibration and image manipulation,to establish the track point division,removal of outliers,three B splines curve fitting the theory of this algorithm,and based on the theory of the algorithm to build a software test platform for tracking trajectory.The tracking trajectory in different welding conditions and workpiece shapes is verified,proving that it is highly adaptable to different scenarios,and the overall average deviation of the experimental trajectory is about 0.26 mm,which fulfills the requirement of actual welding.