Track: Artificial Intelligence

Abstract

Path planning in mobile robots is important since its performance can significantly affect the utilization of robots. Thus we propose a methodology, ACOIC (Ant colony optimization with the influence of critical obstacle), that utilizes the influence values propagated by critical obstacles as the initial pheromones and initial transition probabilities in ACO. Through this approach, we can enhance the traditional ACO by leading ants toward the preferable direction rather than considering all directions in the same weight. Thus the ants are able to reach the goal efficiently without wandering the regions since the optimal path can be obtained proximal to the critical obstacles. In experiment, we implemented the ACOIC and ACO in 3 different maps in terms of the number and shape of the obstacles in order to see if any differences in performance between those two methods exist. As a result, ACOIC was more capable than ACO for generating an optimal path efficiently.