Using Traffic Information for Time-Dependent Vehicle Routing

In this paper we present an experimental evaluation of an algorithm for time-dependent vehicle routing problems using real-world traffic information. Vehicle routing problems (VRPs) have been extensively studied in the literature, mostly with constant travel times. In the time-dependent case travel times depend on the time of the day. For vehicle routing purposes travel time information is provided as time-dependent distance matrices, which are calculated using Dijkstra's label-setting algorithm. Therefore, Dijkstra's algorithm was adopted to integrate time-dependent travel times and to use efficient data structures in order to minimize its runtimes. The computational performance of the implementation was tested with the street network of Vienna and shows promising results. To solve the time-dependent vehicle routing problem with time windows (TD-VRPTW) a Variable Neighborhood Search (VNS) algorithm was applied. The experiment shows, that knowing time-dependent travel times during the tour planning process, it improves the solution quality of the resulting tours significantly.