Multiagent based model for urban traffic congestion measuring

Abstract

In order to measure the urban traffic congestion, a multiagent system that consists of several locally operating agents each achieving a specific task is proposed. The traffic light patterns are calculated by considering different constraints so as to minimize an evaluation criterion. The proposed system aims at managing the gradual congestion of the traffic. As one section of the intersection becomes more congested, the duration of the green traffic lights of the other sections are reduced by the specific agent while the global state of the traffic in the intersection is observed by the other agent. Functionality of the proposed system is examined using initial random data that will be improved by the system.