Vision Position And Control Of Plane Work Piece Based On Adept Robot

Machine vision is used more widely with the development of computer, pattern recognition and image processing. As an important method to get the environment data, the machine vision could improve robotic flexibility and autonomy. It is the very important technology that how industrial robot gets the feature of the work piece fast and recognizes the type of work piece correctly, and judge the position and posture of the work piece in the applications. So the paper targets in the vision position and the control of industrial robot, especially the technology of real-time match recognition and positioning in detail. The main research task follows:(1) The common camera calibration methods used in computer vision are studied and compared. Because of the distance between the work piece and the camera is a constant value during the running time, a nonlinear calibration method that could reduce the camera distort and enhance the precision of inspection is demonstrated in this vision system. The experiments are executed to test the calibration precision finally.(2) The median filtering and histogram equalization are used to preprocess the captured original image in order to improve the image quality. Canny algorithm is introduced to get edge features which show the correctness of positioning and the singleness of response after comparing Canny algorithm to other edge detection algorithms.(3) An improved genetic algorithm and work piece recognition with Hausdorff distance is presented based on the traditional image match algorithm in this paper. The algorithm uses work piece edge feature as match feature and corrected Hausdorff distance as the standard of target similarity measurement, uses genetic algorithm to search the best piece to match. Then, the work piece match recognition is accomplished in distance transformation. The test result shows that this method could find the piece in the motion of movement, rotation and changed scale usefully.(4) By using the recognition and position methods presented above and the vision system that consists of Adept S650 robot, CCD camera, image capture card, motion controller and PC, grabbing work piece experiment is finished.