Design of Vision System for Recognition and Tracking of Fillet Welds of Fan Blade
Industrial fans have the characteristics of complex impeller structure and blade profile,and high machining accuracy requirements.So it has always been a difficult problem to realize automatic welding of blades.A multi-axis linkage robot fan blade welding workstation composed of a robot body,a moving guide and a positioner was designed,which can realize high-precision automatic welding by scanning and locating the weld seam position by laser vision sensor.A new camera calibration method was used to calculate the positions of calibration points on the special calibration board in the image coordinate system and the robot coordinate system to obtain the calibration matrix.LoG filtering and morphological operation image processing algorithms were written to eliminate reflected light and noise points,geometric center method and improved gray centroid method were used to obtain sub-pixel laser stripe center points;straight lines were fitted through RANSAC algorithm and the feature points of the weld were obtained according to the straight line intersection method.Finally,the three-dimensional coordinates of the feature points were obtained according to the hand-eye calibration matrix,so as to guide the robot to realize automatic welding.The welding test results show that the weld seam is well formed without obvious defects,and the position error of the weld is controlled within±1.5 mm,which meets the actual welding process requirements.
fan impellerautomatic weldingweld recognitioncamera calibrationimage processing
NETL
NSTL
万方数据
