An Eye-to-Hand Visual Servoing Controller of 3-DOF Robot Using Feature Matching Technology

Abstract

In robotic control sectors, the image-based visual servo method is an advanced approach used to control the robot to the desired position by using pixels from cameras. We can control the robot flexibly to achieve high quality when the joint variables have little backflash. Applications of this method can be classified into two groups: eye-in-hand and eye-to-hand ones. In this paper, we employ the eye-to-hand method for position tracking control problems of robotic manipulators. Kinematics of the robot and camera are first derived based on homogenous transformation relation and pinhole models. The control technique uses the normalized difference between the pixel observed from the camera and the desired pixel to result in proper control signals throughout a flexible activation gain. The feasibility of this method will be demonstrated through comparative simulation results.