Visual attractiveness in vehicle routing through bi-objective optimization
Vehicle Routing Problem, Visual Attractiveness, Clustering.
In this thesis we approach a vehicle routing problem where the route distribution system must consider both its effective cost and its visual attractiveness. Clustering methods are in principle not designed for the Vehicle Routing Problem, but when used, they can provide visually attractive and possibly cost-effective solutions. So, our proposal is to work in an integrated way in a bi-objective method, which are the route cost minimization and the optimization of a grouping criterion, thus making customers better partitioned in different routes. For this we use a multi-objective evolutionary algorithm based on non-dominance ordering, in order to approximate its Pareto Frontier. We show through computational experiments that our model is capable of generating solutions for vehicle routing that have a low cost and at the same time are visually attractive according to the metrics proposed in the literature. Furthermore, the model was tested with a group of instances based on data from a real road network.