Ant colony system for a problem in reverse logistic

Abstract

Distribution, redistribution, recycling and repacking have become an important issue in logistic planning duringthe last decades. While keeping operational cost as low as possible still the main goal for logistic planners, other aspectssuch as recycling are getting more attention from industry. In this article the well known Ant Colony System (ACS), a bioinspiredalgorithm, is implemented to solve a problem arising in Reverse Logistic namely Vehicle Routing Problem withSimultaneous Delivery and Pickup (VRPSDP). To solve this problem we need to find the optimal set of paths that meet, atthe same time, customer delivery and pickup demands. In order to solve this problem, our ACS implementation makes useof a strategy that mimics the effect of the pheromone in the natural Ants behaviour. To do that, each vehicle is viewed asan individual agent (ant) and consequently its behaviour is driven by pheromone strategy, i.e. it tends to choose the routefor which the pheromone level is higher. Results show that our ACS implementation provides good quality solutionswithin an acceptable time. Furthermore, obtained solutions are quite competitive when compared to other stochastictechniques previously studied in literature.