Arc Welding Robot Seam Correction System Based On Structured-light Vision

The automatization and intelligence of welding process is the development tendency of robot welding technology.The positioning accuracy of the weld is the main factor affecting the automatic welding of robot.The traditional teaching and playing welding robot is not suitable for automatic welding of non-standard container parts.An arc welding robot seam correction system based on structured-light vision is designed and studies.Aiming at the non-standard container parts with poor positioning accuracy,the weld trajectory is corrected by offline and online correction methods to automate the welding process and improve the efficiency of welding.The measurement principle of the arc welding robot vision system is analyzed.The calibration of the tool and the structure light sensor are completed.The hand-eye calibration and the light plane calibration of the structure are completed by the two-step method and the cross-ratio invariance respectively.An automatic calibration method for the structured light vision system is adopted.Combined with the common types of weld joints,a seam feature recognition method is adopted with piecewise linear fitting strategy based on the laser center line which is extracted by active contour model method.Special conditions such as butt weld and pre-welding spot are dealt with special methods.A scanning method for detecting the initial welding position is designed.The offline correction method is proposed considering the structure of workpiece.A rapid positioning method is proposed for the circular weld and the relevant experiments are carried out.To solve the problem of trajectory deviation caused by the size error and shape error of the workpiece,a trajectory matching method with global matching then piecewise matching is proposed.Experiments for space multiline weld and straight line and curve weld are conducted.The online correction control method is proposed.Online correction algorithm and data storage method are designed based on sample period.Kalman filter is used to smooth the detected seam points.A fuzzy-PI double-layer control method is adopted to drive the robot to move faster and more accurately.Experiments for space curve weld and space intersecting line weld are well conducted.The correction error is within 0.3mm.