Most of the models for vehicle routing reported in the literature assume constant travel times. Clearly, ignoring the fact that the travel time between two locations does not depend only on the distance traveled, but on many other factors including the time of the day, impact the application of these models to real-world problems. In this paper, we present a model based on time-dependent travel speeds which satisfies the ''first-in-first-out'' property. An experimental evaluation of the proposed model is performed in a static and a dynamic setting, using a parallel tabu search heuristic. It is shown that the time-dependent model provides substantial improvements over a model based on fixed travel times.

A n important, but seldom investigated, issue in the field of dynamic vehicle routing and dispatching is how to exploit information about future events to improve decision making. In this paper, we address this issue in a real-time setting with a strategy based on probabilistic knowledge about future request arrivals to better manage the fleet of vehicles. More precisely, the new strategy introduces dummy customers (representing forecasted requests) in vehicle routes to provide a good coverage of the territory. This strategy is assessed through computational experiments performed in a simulated environment.

Recent technological advances in communication systems now allow the exploitation of realtime information for dynamic vehicle routing and scheduling. It is possible, in particular, to consider diverting a vehicle away from its current destination in response toIn the past few years, there has been a rapid growth in communication and information technologies (e.g., global positioning satellites, cellular phones, geographic information systems, geosynchronous satellite-based systems, etc.). These recent advances afford opportunities for using real-time information to enhance the performance of decision systems in the area of vehicle routing. According to the fraction of requests that are known in advance, vehicle routing problems may be classified as static or dynamic. In the static case, all data are known before the routes are constructed and do not change afterward (e.g., location of transportation requests, demand, etc.). However, in the dynamic case, all or a fraction of all requests are revealed as the routes are executed. Hence, dispatchers are forced to react to events that occur in real time, such as new service requests, unexpected delays, accidents, etc. Numerous examples of dynamic vehicle routing and dispatching problems may be found in practice, like ambulance or police services, courier services, diala-ride problems (e.g., transportation-on-demand for the handicapped) and many others.Due to the new information technologies mentioned above, real-time information is now available at lower costs. This explains the new interest that dynamic vehicle routing problems have gained recently. However, there is still a lack of methodologies that can efficiently solve dynamic vehicle routing problems through a judicious integration of realtime information.In dynamic vehicle routing problems, one potential use of real-time information is to divert a vehicle away from its current destination to serve a request that just occurred in the vicinity of its current position. In this work, we propose an approach for the dynamic assignment of new requests, which includes diversion, and we examine different ways of implementing it. An empirical evaluation is performed within the tabu search heuristic reported in GENDREAU et al. (1996b).The problem considered is motivated from a courier service application found in the local operations of long-distance shipping services, where the mail is collected at different customers' locations and brought back to a central office for further processing and shipping. This problem belongs to the class of pick up (or delivery) only problems where a set of requests must be transported to (or from) a central depot. The goal is to design a set of minimum cost routes, originating and ending at the central depot, to satisfy the transportation requests. In a dynamic context, each new request is inserted in real time in the current set of planned routes, where a planned route is the sequence of requests that have been assigned to a vehicle but have not been serviced yet.The paper is organized a...

No abstract