Solving Robot Path Planning Problem Using Ant Colony Optimisation (ACO) App

Abstract

Learning is a complex cognitive process; thus, the algorithms that can simulate learning are also complex. The complexity is due to the fact that little is known about the learning process that can be simulated in a machine. In this study two methods have been chosen to navigate a simulated robot to a target point; namely, Ants Colony Optimisation (ACO) and the Fuzzy Approach. The focus of this paper is primarily the ACO method and the Fuzzy Approach is used as a comparative benchmark. Three scenarios were designed: the Big Hall, the Wall Following and the Volcano Challenge. These experimental scenarios represent the respective navigation frameworks found in the literature used to test learning algorithms. The results indicate that the ACO’s performance is inferior to the Fuzzy approach; justification for this has been discussed in relation to previous research in this area. Some future work to investigate this phenomenon further and improve the performance of the ACO algorithm is also presented.