Design Of Manipulator Visual Servo System Based On DM6437Platform

As the core technology of today’s industrial automation field，robotics technologyis now leading the wave of industrial development. In the field of robotics technology,robot vision is becoming its key part of intelligent process. Compared to other sensoryinformation, visual sense has a greater amount of information and wider applicationprospects. By adding visual information into the control system, robot visual servotechnology improves the level of automation and intelligence and gains greatsignificance. In this paper, the robot visual servo system is built based on RM-501manipulator and the manipulator’s move is guided by visual information.First of all, in manipulator motion control section, the forward kinematicsmathematical model and the inverse kinematics mathematical model are created bymodified D-H method and verified by robotic toolbox in MATLAB.Secondly, determine the joint motors’ model parameters through experiments.Photoelectric encoder is selected as the feedback device of position and velocityinformation. Double-loop control systems of manipulator’s joints are also built andtested in RTW environment, which also verifies the rationality of the system modeling.Thirdly, in the image processing section, the video and image of moving target iscaptured by DM6437processing platform. Firstly, the image feature is highlighted andthe effect of noise is reduced by conducting the image preprocessing. Secondly, byusing sobel operator, the edge detection and feature extraction are completed. Finally,according to the extracted features, the center of moving target is marked and thecoordinate of which is introduced into the control system to be the overall inputinstruction.At last, a visual controller of Fuzzy network is designed. The rationality of thesystem is tested and the overall performance of the system is analyzed in the MATLABenvironment.