A real-time vehicle routing model for a courier service problem

The information needed to solve a vehicle routing problem may be not completely known ahead of time. Rather it may be dynamically revealed as time goes on. We consider a dynamic vehicle routing problem faced by a courier company where customer requests with service time windows arrive and have to be serviced on real time by a fleet of vehicles in movement. Differently from other dynamic routing problems motivated by the same courier service, we consider both pick-up and delivery requests and assume that customer requests cannot be refused but can be postponed to future shifts. A heuristic algorithm based on local search is proposed for the problem together with an illustrative example. Experimental analysis is in progress.