Cooperative co-evolutionary memetic algorithm for pickup and delivery problem with time windows

Solving rich vehicle routing problems has become an important research avenue due to a plethora of their practical applications. Such discrete optimization problems commonly deal with multiple aspects of intelligent transportation systems through mapping them into the objectives which should be targeted by the optimization algorithm. In this paper, we introduce the cooperative co-evolutionary memetic algorithm for this task. It benefits from the simultaneous evolution of several subpopulations, each corresponding to a single objective, and from the process of migrating the best individuals across such subpopulations to effectively guide the search process. The experimental study performed over widely-used benchmark test cases indicates that our algorithm significantly outperforms the memetic techniques which tackle each objective separately and those that turn the multi-objective problem into a single-objective one through weighting the optimization criteria.