VNS-based multi-agent approach to the dynamic vehicle routing problem

Abstract

The paper focuses on Dynamic Vehicle Routing Problem (DVRP), where customers’ requests arrive dynamically while the process of planning and execution of the routing plan is running. Typically, the dynamic problem increases the complexity of the problem and introduces new challenges while finding the optimal route plan. The main contribution of the paper is to propose a multi-agent approach to the DVRP with efficient VNS-based procedure to periodic re-optimization of static subproblems, including requests, which have already arrived to the system. The results of evaluation of the proposed approach confirmed its practical ability do simulate and efficient solve the DVRP.