An Empirical Study for Fitness Function Selection in Fuzzy Logic Controllers for Mobile Robot Navigation

Abstract

Fuzzy logic is widely used for mobile robot navigation. The main draw back of this approach is the ad hoc design of the controllers used. A popular method for the optimization of fuzzy logic controllers for the navigation of mobile robots is the use of genetic algorithms. An issue, in this procedure is the selection of the fitness function for the improvement of the behavior of a pre-designed controller. We analyze the factors that influence the evolution of the fuzzy controller based on the fitness function used and present some preliminary results. In order to validate our approach a test bed has been developed based in a low cost robot