Research on Robot Visual Servo Control Based on Image Identification

摘要

The research is mainly about robot system, DPS image processing system and Jacobian matrix. The robot in this study is a plenary robot with two rotary joints. It can only move in two directions, x and y. The joints are powered by Panasonic ac servo motor. The CCD camera is installed at the end of the second arm. The math model of the robot visual servo system based on image was set up here. By avoiding the transformation of three-dimension coordinate, Jacobian matrix and its inverse matrix that show the relationship between the coordinate variation and the angle variation of the joints were set up. The two matrices were also used in the visual servo based on image. DSP image processing system processes the video image of object taken by CCD camera. It first transforms the image into gray level image. Then it gets the boundary of the round object after going through the processes of image binaryzation, median filtering and boundary extraction algorithm. Barycentric coordinates will be evaluated through point Hough transformation. The outcome will be sent to master control program of the Industrial Personal Computer through serial ports and thus the visual servo circulation is formed.