Dynamic Vehicle Routing Problem with real-time time-dependent travel times

The real-time decision problems play a more and more important role in many areas such as the logistic distribution. This paper is focused on the Dynamic Vehicle Routing Problem with Time Windows (DVRPTW) in the context of real-time traffic information. The travel times are based on a time-dependent model in which the travel speeds are step functions. This model is enriched with an exponential smoothing function able to calculate the forecasted speed. This new model of travel times is used by on-line genetic algorithm for the route planning. The validation of this approach on benchmarks shows that the real-time traffic information can improve performances of the route planning.