Online K-means based heuristic for the dynamic pickup and delivery problem solving

We consider the Dynamic Pick-up and Delivery problem with Time Windows (DPDPTW). It is about goods to be carried from a pickup to a delivery location. Real world applications of DPDPTW arise, for example, in courier operations and door-to-door transportation services. Our purpose is to solve efficiently the DPDPTW, taking into account the real time customer requests for transportation services to produce a solution that minimizes the operational costs and fulfills the problem constraints. For this aim, we propose an online K-means based heuristic. The requests are processed one per one, clustered into groups using the K-means algorithm. Routing and scheduling are performed using a dynamic programming algorithm. Computational results show that the proposed method yields good solutions compared to literature construction and insertion methods.