A hybrid population seeding technique based Genetic Algorithm for stochastic Multiple Depot Vehicle Routing Problem

Abstract

Vehicle Routing Problem (VRP) has wide applications in logistics and supply chain management and thus is one of the widely studied problems in the field of Operational Research. It is also a NP-hard combinatorial optimization problem and many different kinds of algorithms and techniques have been proposed to solve VRP. There are many types of VRP and this paper concentrates on two variants: Multiple-Depot Vehicle Routing Problem (MDVRP) and Stochastic Vehicle Routing problem (SVRP). While both MDVRP and SVRP enjoy wide popularity in literature, a combination of these two is not yet explored. The objective of this paper is to solve for MDVRP with stochastic travel times using a metaheuristic procedure in Evolutionary Computation called Genetic Algorithms (GA). A new hybrid population seeding technique is proposed for generating feasible solutions in the initial population. A randomized initial population generation technique is used to compare with the proposed hybrid population seeding technique and the results are compared. The results clearly conclude that the hybrid population seeding technique clearly yields better solutions in terms of time needed and distance travelled to serve the customers.