ANFIS-based aided stabilization control and image tracking for mobile robot

Abstract

Vision systems are widely used on mobile robot systems. When the mobile robot travels on a bumpy road, a fixed camera cannot track object stably. Specifically, when the mobile robot is searching and/or rescuing in a valley or a dried river bed, how to achieve stable image tracking is a very important issue. This paper devotes to develop a stabilization control scheme for a mobile robot system running on different road profiles and rough terrain while aiming at a target. Firstly, the image recognition algorithm and the outer-loop PI controller for a two-axis camera system have been developed. A simulation system composed of two-axis camera system as well as 3-DOF rotating platform has been set up to imitate a physical dynamic of a mobile robot system. In addition, we have succeeded in building up a three-axis to two-axis ANFIS transformation scheme from a data set of measurements. For demonstrating the method, the data obtained from a measurement of a physical road profile and rough terrain that a vehicle experienced is fed to the system. The targeting error of the two-axis camera targeting at the object under the disturbance of road profile and rough terrain is in a circular region with diameter 20 cm, and is within an acceptable range.