Calibration Algorithm of Mobile Robot Vision Camera

Abstract

This paper proposes a method that can calibrate the camera installed in a mobile robot. This method works by using the principle of camera's perspective projection and the pinhole imaging model, and obtain static external camera parameters through a three wire calibration method. In consideration of robot's bumpy moving condition, a camera height dynamic compensation algorithm which depends on the characteristics of parallel lines is put forward. This camera height dynamic compensation algorithm is based on the inherent characteristics of using two parallel lines to conduct a dynamic compensation for the heights of a robot vision camera. A comparison between the calibration error of this algorithm with that of other calibration algorithms shows a clear advantage of this method over the others. According to the result of an undulating road experiment, the height values obtained by using this algorithm are closer to the actual heights. Analysis and experiment results show that this calibration compensation algorithm can significantly reduce calibration error caused by undulating road.