Machine Learning Techniques in Vehicle Routing Problem

Abstract

The globalization of trade, urbanization, and the international division of the labor are the main factors, that have led to an increasing need for transportation. In fact, with the opening of different economic markets and the interdependence between countries through the supply and demand of goods. Companies around the world are increasingly facing globalization effects. Specifically, they must satisfy their customers and improve their performance by optimizing each sub-process in supply chain process Vehicle routing problem one of the important combinatorial problems for goods distribution as well as for passengers’ transportation. Over time, many variants of this problem were studied and several solving approaches are proposed to reach optimal solutions with exact methods or near optimal ones using approximate algorithms (heuristic and metaheuristic) . Recently, Machine learning techniques are explored to find the best solutions of VRP problem with high performance . In this review, we present the classical formulation of the vehicle routing problem as well as its most prominent variants. We provide an overview of the most significant machine learning concepts and techniques used in the literature for solving VRPs. To classify the papers, we followed the research methodology proposed by Mayring and we finally discuss the findings, analyze the results, and compare each machine learning techniques applied to this problem with other techniques.