Author(s)

J. W. Joubert

Abstract

The integration of multiple constraints of the Vehicle Routing Problem (VRP) variants is computationally expensive. Although vehicle routing problems have been well researched, variants are typically treated in isolation, whereas industry requires integrated solutions. Solution algorithms are also tested using benchmark data that are questionable, and that do not represent typical applications. The paper proposes an approach that solves a problem by analyzing its environment through cluster analysis, chooses an appropriate solution strategy, and tests the results in an attempt to learn for the purposes of improved future decisions. Keywords: vehicle routing, VRP, heuristics, metaheuristics, artificial intelligence, learning. 1 Introduction Vehicle routing and scheduling problems are well-researched in the field of Operations Research. The main objective of these types of problems are to minimize the distribution costs for individual carriers. Given the complexity of the type of problem, extensive research has been conducted to develop exact and heuristic solution techniques for urban distribution problems. The Vehicle Routing Problem (VRP) can be described as the problem of assigning optimal delivery or collection routes from a depot to a number of geographically distributed customers, subject to side constraints. In its basic form, the VRP can be defined with G = (V,E) being a directed graph where V = {v0, v1, . . . , vn} is a set of vertices representing customers, and with v0 representing the depot where m identical vehicles, each with capacity Q, are located [1]. E = {(vi, vj)|vi, vj ∈ V, i = j} is the edge set connecting the vertices. Each vertex i, except for the depot (V \vo),

Keywords

vehicle routing, VRP, heuristics, metaheuristics, artificial intelligence, learning.