Abstract:The vehicle routing problem with simultaneous pickups and deliveries (VRP_SPD) is a variant of the classical vehicle routing problem (VRP) where clients may require simultaneous pickups and deliveries service. An improved genetic algorithm was proposed to deal with pickups and deliveries in an integrated manner, instead of traditional insert-based heuristics. An enhanced edge recombination crossover was designed, which could keep the inheritance of edge-adjacency relationship between the two parent routings and thus result in an improved performance. A self-adaptation strategy was applied to control parameters which achieved a better robustness. Numerical experiments and simulation have been conducted which demonstrate that the algorithm can obtain even better results compared with the heuristic method commonly in use, and demonstrates a good robustness under the random environment.

Improved Genetic Algorithm for Vehicle Routing Problem with Simultaneous Pickups and Deliveries