Fuzzy Counter Ant Algorithm for Maze Problem

Abstract

This effort explores the effectiveness of adding a layer of fuzzy logic to a group of swarming multi agent robots for exploration and exploitation of an unknown obstacle rich environment represented by a 2D maze problem. The generalized maze problem has been considered as an interesting test bed by various researchers in AI and neural networks. Using a cooperative multi agent robot system reduces the convergence time considerably as compared to a single agent. For the multi agent case, a robust and effective decision making technique is required that prevents a robot from moving to a region already explored by some other robot. In this paper, we present a counter ant algorithm (modified ant colony optimization algorithm) based on a fuzzy inference system which enables multiple agents in path planning along the unexplored regions of a maze in order to find a solution rapidly. Simulation results demonstrate the effectiveness of this approach.