Research On Online Inspection Technology Of Robot Automatic Drilling Based On Vision

With the improving of the intelligent machinery manufacturing,aviation industry developing rapidly,the machining quality and efficiency of modern aviation manufacturing industry put forward higher requirements,and the plane as the aviation industry,the most representative products realize aircraft parts,assembly and testing of the automation system in the process of hole,is the effective way to improve the efficiency of aircraft manufacturing and quality.The traditional manual detection method used in aviation industry is slow,inefficient and easy to cause the situation of missing inspection and wrong inspection.In order to improve the efficiency of machining and assembly,it is very important to build a real-time and efficient on-line inspection system to improve the automatic hole making technology of robot.Based on the machine vision method,this paper studies the robot automatic hole making online detection system.The main research contents are as follows:Firstly,the overall scheme of the robot automatic hole making online detection system was designed.Eye-in-hand hand-eye calibration system configuration was adopted to establish the robot system coordinate system,and coordinate transformation was carried out for the robot coordinate system,workpiece coordinate system and camera coordinate system to determine the internal and external parameters of the camera.The hole position was identified by the template matching method,and the hole edge was fitted by the Hough circle detection algorithm,and the hole size was detected according to the hand-eye calibration results.Secondly,the paper studies the inlet and outlet defects of Carbon Fibre Reinforced Plastics(CFRP).The images of the inlet and outlet defects of CFRP are pre-processed by means of weighted mean grayscale method,adaptive mixed filter noise reduction method and logarithmic transformation image enhancement method.The OTSU adaptive threshold selection algorithm was used to determine the ROI region of the image,and the open closure algorithm was used to fill the holes and complete the two value images of the divided defect regions.Canny operator is used to obtain pixel-level edge of hole surface defect,so as to accurately locate the edge and obtain complete defect contour information.Thirdly,a Multi Layer Perceptron(MLP)based defect classifier was designed to realize defect classification.The recognition accuracy and detection time of MLP classifier were tested,and the accuracy of classification was verified through experiments.MLP classifier was applied in the robot automatic hole making online detection system.Finally,the robot system of automatic hole online detection system hardware choice and integrate it to the end of the multi-functional actuators using MFC application module in Visual Studio 2015 development tools,combined with Open CV image processing library,Halcon image processing library development programming with mixed on-line detection module and offline testing image processing module,complete automatic control hole online test system software and hardware development of robot.