The alliance between optimization and multi-agent system for the management of the dynamic carpooling

Abstract

Today, there are several studies that revolve around dynamic carpooling. However, there is a big handicap, due to the problems high complexity, concerning the way to make the process perform efficiently. To address these gaps, we introduce a decomposition process in order to subdivide the global problem into several sub-problems with a reasonable research space. Indeed, we propose to break geographical areas (global problem) into several distinct zones (sub-problem) which each zone is controlled by an agent with an optimized behavior. Therefore, we propose the original alliance between optimization and a multi agent concept to perform parallel Optimized Assignment of Vehicles to users queries. This alliance is characterized by a metaheuristic approach based on a Multi-criterion Tabu Search implemented in the heart of the agent in order to optimize partial requests process which is performed locally in its zone. Moreover, we introduce several agents which are endowed by an evaluator behavior based on the Choquet Integral to evaluate the best solution taking into consideration the interactions among criteria. Finally, to test the validity of the proposed model, some simulation results will be presented.