Real-Time Control Of Welding Seam Tracking Based On Vision Sensor With Structural Stripe

This thesis focuses on the study of welding seam tracking system based on laser structural light. Autonomous tracking method is used to track real welding seam in this system. Bireflect laser vision sensor, Data acquisition card, Motoman SSA2000 robot, Motocom32 communication software and LabVIEW are used to form software platform.Matlab calibration tool kit is used to realize high-accuracy calibration of camera inner parameter. Calculation error of laser plane is decreased by using symmetrical rotation method. A best hand-eye relation judging rule in hand-eye relation calibration is rasied. This rule is used to calculate robot hand-eye relation. Matrix eigenvalue solving method and Newton-rafer's iterative method were used to calculate inverse kinematics of nonspherical wrist robot.Welding seam modeling method which can be used to compute position and direction information of welding seam is raised. This thesis uses the forward distance between the vision sensor and the welding gun to generate a series of TCP points corresponding to the welding seam path points. To realize six-dimensional welding seam tracking, author call Motocom32 dll(dynamic link library) to create and upload welding task to robot. This method is effective to eliminate look-ahead error. It can track welding seam precisely in theory.Tracking experiment of straight line and curve welding seam indicates that this system has good position tracking ability and pose tracking ability. Tracking precision of system is within 0.5mm.