An agent-based guided local search for the capacited vehicle routing problem

Abstract

The main contribution of the paper is to propose and validate a new agent-based method of solving instances of the Capacited Vehicle Routing Problem (CVRP). The approach adopts a guided local search metaheuristic and combines it with an asynchronous team (A-Team) concept. A-Team assumes that a collection of software agents, each representing a particular problem solving method, cooperate to solve a problem by dynamically evolving a population of solutions. In suggested implementation each software agent carries out a guided local search. The paper contains the CVRP formulation, an overview of the dedicated multi-agent framework and a description of the proposed implementation for solving CVRP. The approach is validated experimentally and results of computational experiment are included in the final part of the paper. © 2011 Springer-Verlag.