Structured-light Correction System On Arc Welding Robot For Nonstandard Vessel

Nonstandard vessel is common in different products of the dyeing industry, aerospace industry and transportation sector, etc. The automatic welding process is crucial to the quality of those product. The low seam positioning accuracy of nonstandard vessels greatly hinders the automated welding. To improve both the welding quality and the efficiency, a structured-light correction system on arc welding robot is designed both in the offline and online case, while ensuring welding angles and avoiding collision.In this paper, the structured light sensor used for seam detection is designed based on the analysis of welding material, type and noise and the derivation of structured light measuring principle. A culling algorithm is devised based on analysis of join skeleton and non-join skeleton, which is used for initial laser line recognition under the presence of specular noise. Then the B-Snake model is studied to recognize the laser line effectively under the noise of the GMAW welding process, such as welding arc, patter and smoke.After the calibration both of robot tools and camera, a two step calibration method based on hand-eye calibration and light plane calibration is designed. Based on piecewise seam trajectory and cubic Bezier curve, an improved iterative closest point registration algorithm is proposed to conduct accurate registration between seam of new workpieces and the initial teaching trajectory. The offline correction producing new teaching points located in the new seam combined with the original posture information achieving improved welding quality and collision avoidance. Two dimensionless evaluation method of seam registration is proposed.Correction system's architectures and operating procedures are devised. The online correction software architecture is lay out based on Yaskawa arc welding robot. Online correction algorithm is designed and simulated based on the motion interpolation cycle. Based on the TCP protocol and the ring storage area, sem position data transmission between main host computer and the robot controller is devised. Finally, experiments of online welding correction are carried out. The error is within 0.5mm.