Track model regression using genetic expression programming for visual-based path-following of mobile robots

Path-following is an important task in the applications of autonomous robots. This paper presents a new control method for the control of visual path-following of mobile robots. Visual path-following of mobile robots is rephrased as a regression problem which takes the visual positions of tracks as the input and produces control signals as output. In this paper, the technique of Gene Expression Programming (GEP) is employed for discovering such regression models. With the regression models, path-following can be done with simpler control rules with less processing time. Our proposed method is implemented and integrated with a two-wheeled service robot carrying a simple CCD camera and is validated for visual path-following in real environments. The experimental results show that the proposed method is able to produce simple and precise control rules and improves the robot's performance for path-following.