A Hybrid Ant Colony Optimisation Technique for Dynamic Vehicle Routing

Abstract. This paper is concerned with a dynamic vehicle routing problem. The problem is dynamic in the sense that the time it will take to traverse each edge is uncertain. The problem is expressed as a bi-criterion optimisation with the mutually exclusive aims of minimising both the total mean transit time and the total variance in transit time. In this paper we introduce a hybrid dynamic programming - ant colony optimisation technique to solve this problem. The hybrid technique uses the principles of dynamic programming to first solve simple problems using ACO (routing from each adjacent node to the end node), and then builds on this to eventually provide solutions (i.e. Pareto fronts) for routing between each node in the network and the destination node. However, the hybrid technique updates the pheromone concentrations only along the first edge visited by each ant. As a result it is shown to provide the overall solution in quicker time than an established bi-criterion ACO techniq...

• Type: Conference
• Year: 2004
• Where: GECCO
• Authors: Darren M. Chitty, Marcel L. Hernandez
• Keywords: Destination Node, GECCO 2004, Total Mean Transit, Transit Time