Intelligent Agents For Dynamic Vehicle Routing

Introduction

Intelligent Agents have enormous scope in the Transportation domain because of the domain's dynamism (rapidly changing traffic conditions), multi-modality (people and goods travel across land, sea and air), and its spread over large geographical area. Agent characteristics like autonomy, collaboration with other agents, and most of all, mobility will open up new applications that may be difficult to achieve with traditional client/server technology. But, currently there are very few fielded agent applications in the Transportation domain. This paper discusses a specific application of Dynamic Vehicle Routing, the incorporation of agents technology in it, and the advantages of agents over client/server architecture.

A Dynamic Vehicle Routing Application

Consider an in-vehicle application that provides advice to the driver on the best route to take for his trip. As the driver gets into his vehicle, he will specify his destination to his In-Vehicle Unit (IVU). The system plans the shortest route using standard shortest path algorithms based on some norms. Once a tentative route is planned, the application sends out intelligent mobile agents to traffic information providers (or `Places' in agent terminology) along the planned route. The traffic information providers have information about current and historical traffic conditions and expected travel times in that region. The traffic information providers may gather these information from the traffic sensors in their region. If the conditions are expected to differ from the norm, the agents will send information back to the IVU for reconsideration. The driver in consultation with IVU may change the route, in which case new mobile agents may be sent to the traffic information providers along the new route or the old ones may be redeployed. The process continues until the destination is reached. In short, mobile agents are always a few steps ahead of the driver to monitor traffic conditions for him and suggest him better routes.

In our formal paper, we would elaborate on the specific ways in which intelligent and mobile agent technologies can be incorporated in the system. We would also discuss the many advantages such an approach has over the client/server approach.

Conclusion

The Transportation domain by its very nature of dynamism and diversity is highly suitable for employing agents paradigm. The agents approach has many real advantages over the client/server approach in terms of autonomy, mobility, better response and customizability. Dynamic Vehicle Routing is one such application that the paper addresses.

Ranganatha Sitaram <ranga@iti.gov.sg>
Sek-Wah Tan <tsw@iti.gov.sg>

Information Technology Institute

11 Science Park Road

Singapore Science Park II

Singapore 117685