In this article, the single capacitated vehicle routing problem with time windows and uncertain demands is studied. Having a set of customers whose actual demand is not known in advance, needs to be serviced. The goal of the problem is to find a set of routes with the lowest total travel distance and tardiness time, subject to vehicle capacity and time window constraints. Two uncertainty types can be distinguished in the literature: random and epistemic uncertainties. Because several studies focalized upon the random aspect of uncertainty, the article proposes to tackle the problem by considering dominance relations to handle epistemic uncertainty in the objective functions. Further, an epistemic multi-objective local search-based approach is proposed for studying the behavior of such a representation of demands on benchmark instances generated following a standard generator available in the literature. Finally, the results achieved by the proposed method using epistemic representation are compared to those reached by a deterministic version. Encouraging results have been obtained.
A Hybrid Multi-objective Genetic Algorithm for Bi-objective Time Window Assignment Vehicle Routing Problem
