Current robot control techniques are limited in the main to the teaching method. In this study, construct a robot vision system using end-effector mounted C.C.D camera and present a calibration method for its automatic teaching system. Machine vision is becoming increasingly practical for use in industrial control. The emergence of the industrial robot, and its potential for performing heretofore manually applied industrial processes, has created the need for development of human-like sensing capabilities.
The proposed calibration method is divided into three steps. In the first step, the intrinsic parameters of camera system are recovered from a double image of a cube using orthogonal properties of vanishing points with respect to camera center. In the second step, determine pose variables representing the orientation and position of camera with respect to the scene using one known length of the scene. In the third step, find the extrinsic parameters, the position and orientation of the camera with respect to the robot tool, by moving the robot twice and observing the resulting motion of the sensor. Test results of real experiments are presented and compared with the real datas and analyzed.