Optimizations for Multiple Collective Sources in Delivery Systems

Abstract

In this paper we investigate a new subset of delivery problems where the destinations are all to be delivered from one or more sources so that the total distance is minimized. For example, food is delivered for the customers who place orders from one or more restaurants. For one source, we propose and compare three greedy algorithms namely nearest neighbor first (NNF), polar angle sweep (PAS), and distance sweep (DS). For multiple sources, each destination is from a specific source, thus requiring that a destination must be visited after its source. We give an optimization algorithm called "collect all then distribute" (CATD). We conducted comprehensive experiments based on various synthesized data sets and compared the accuracy and runtime complexity of the proposed algorithms. Our conclusion is that the NNF and CATD algorithms have clear advantages over other alternatives.