GRACE: A generational randomized ACO for the multi-objective shortest path problem

Abstract

The Multi-objective Shortest Path Problem (MSP) is a widely studied NP-Hard problem. A few exact algorithms were already proposed to solve this problem, however none is able to solve large instances with three or more objectives. Recently, some metaheuristics have been proposed for the MSP, but little can be said about their efficiency regarding each other, since no comparisons among them are presented in the literature. In this paper an Ant Colony Optimization (ACO) algorithm, called GRACE, is proposed for the MSP. The proposed approach is compared to the well-known evolutionary algorithm NSGA-II. Furthermore, GRACE is compared to another ACO algorithm proposed previously for the MSP. Results of a computational experiment with eighteen instances, with three objectives each, show that the proposed approach is able to produce high quality results for the tested instances. © 2011 Springer-Verlag.