In this paper we present a greedy randomized adaptive search procedure (GRASP) for solving a vehicle routing problem (VRP) for package delivery with delivery place selection. The problem can be solved by stepwise optimization, i.e., first selecting delivery sites and then defining routes based on that selection. Alternatively, it can be solved by jointly optimizing delivery site selection and routing. We investigate the effects of stepwise optimization in comparison to joint optimization. The evaluation results show that our proposed stepwise approach, while expectedly producing longer routes than joint approach (by \(4\%\) on average), can provide a solution \(1000\times \) faster than the previous benchmark approach. The proposed procedure is therefore well suited for the dynamic environment of package delivery which is widespread in modern cities as a consequence of e-commerce.
- Greedy randomized adaptive search procedure
- Vehicle routing problem
- Package delivery