The multi-pickup and delivery problem with time windows

This paper investigates the multi-pickup and delivery problem with time windows in which a set of vehicles is used to collect and deliver a set of items defined within client requests. A request is composed of several pickups of different items, followed by a single delivery at the client location. We formally describe, model and solve this rich and new problem in the field of pickup and delivery vehicle routing. We solve the problem exactly via branch-and-bound and heuristically developing a hybrid adaptive large neighborhood search with improvement operations. Several new removal and insertion operators are developed to tackle the special precedence constraints, which can be used in other pickup and delivery problems. Computational results are reported on different types of instances to study the performance of the developed algorithms, highlighting the performance of our heuristic compared to the exact method, and assessing its sensibility to different parameter settings.