On exploring a virtual agent negotiation inspired approach for route guidance in urban traffic networks

Abstract

The traditional route guidance system often provides the same shortest route to different drivers regardless of their different traffic conditions. As a result, many vehicles may rush into the same road segments at the same time that would lead to traffic congestion. Such uncontrolled dispersion of vehicles can be avoided by evenly distributing vehicles along the potential routes. This paper proposes a practical Virtual Agent Negotiation based Route Guidance Approach (VANRGA). In the proposed approach, vehicle agents (VAs) in the local vicinity communicate with each other before the intersections to achieve a real-time and dynamic route selection. Based on the route preference of the drivers and the traffic conditions, the vehicles are distributed on the routes equally, which can avoid the traffic congestion and maximize the utility of the road resources. After presenting the design and implementation methodology of VANRGA, this paper carries out extensive experiments on synthetic and real-world road networks. The experimental results show that compared to the shortest path algorithms, VANRGA offers a 22%-37% decrease in travel time (when traffic demand is below network capacity) and a 15%-18% decrease in travel time (when traffic demand exceeds network capacity).