| With the rapid development of computer science and electronics,imageprocessing, photoelectric technology, and the theory and technology of patternrecognition, the actual research and application of robots visual have been paid moreattention increasingly, and made preliminary achievements in many fields constantly.At present, arc welding robots mainly used in the industrial production lack straincapacity, and are demanded in the stability of operation condition. There is a robotwelding process with all kinds of complicated uncertain interference factors, so theresearch based on visual sensing automatic welding technology to improve theautomation and the intelligence level of welding robot is of great significance.This paper takes the ABB welding robot in Jilin Taihe group co. Ltd. as theresearch object, to achieve its purpose of automatic welding, mainly focused on themachine vision to realize automatic welding, on the basis of the existing arc weldingrobot, designed and implemented two-dimensional automatic welding robotsimulation system based on visual sensing technology guide.This paper designed the hardware and software of the automatic welding system,build the system framework. Basic working principle of the system is: catching theweld image through the visual sensor, processing the image on the computer, gettingthe weld coordinates, and sending the coordinates of the welding to the robotcontroller, the controller realize automatic welding according to the weld coordinateproduction path.First of all, with the reference for related research at home and abroad, thedesign of this study in this paper are the visual guide of automatic welding seamimage acquisition, the hardware constitution and the software process, meanwhile,finishing the comparison and selection of the hardware. Secondly, it does a lot ofresearch on camera calibration method, uses the "two-step method" for cameracalibration, and gives the detailed algorithm process and experimental process. The feasibility of this method is verified by experiment. Then, in view of the acquisitionof image, a complete set of image processing algorithm is studied, including imagepre-processing,(image pre-processing includes image filtering, image enhancement,and binarization) isolated point filtering, edge detection, linear fitting, finallyextracting weld coordinate effectively, at the same time meeting the requirements ofthe rapidity. Finally, this paper write the robots program in RobotStudio software,formulate detailed experimental steps, do the two dimensional robot automaticwelding simulation experiment, analyze the experimental data, and verify thefeasibility of the system. |