Research On Target Recognition And Grabbing Of Bending Robot Based On Machine Vision

The CNC bending machine with assist robot currently used has a trend of replacing the single CNC bending machine or the CNC bending machine with a assistant bracket in front, the efficiency and velocity of the CNC bending machine with assist robot is much higher than the latter two, the sheet metal material of CNC bending machine mostly is wide and thin plate,these plate have different sizes, adopt different positioning tables and affect the automatic production, Currently using a common mechanical structure positioning table exists two questions:One is the identification of plate’ stype;the other is the estimation of the poisition and angle of plate in the drop clip area.To cope with the problems mentioned above, In this paper, robotics and machine vision technology, through the use of machine vision systems to assist in identifying robots CNC bending machine bending of plates correctly. This study includes:Fist of all, First with histogram equalization, median filtering to improve the images obtained from the industrial quality of the camera, After based on edge detection method, extract the needle plate workpiece material characteristics, using median filtering and histogram equalization first image pre-processing, and then use the edge detection method for image feature extraction, through a variety of edge detection operators compared to last Canny operator as part of this feature extraction method.Secondly,Make use of HALCON programming environment to recognize and match the image of parts,it can complete the workpiece positioning by extracting the edge of special points and the center point. Obtained using the Hough transform parts of the boundary point set, find the correct plate boundary, in order to complete the positioning of the workpiece below the foundation.Thirdly,The traditional method is used for calibrating the cameras,then the visual system begins to recognize and locate all kinds of sheet metal in store,the results will be feed back to the control system of the robot and it control manipulator sucker to suck up. Adjust the manipulator according to the object posture of visual location and make it suck up the central location of sheet material.Last,by using the recognition and position methods referrd above and the hardware syetem,which consist of HR50robot, CCD camera,camera shot,and PC,grabbing component experiment is a accomplished.