In this paper we propose the application of intelligent agents in traffic-lights, for the road control of urban transit in the city of Colima, México; using a multi-agent approach for dynamic urban traffic-lights system coordination that relies on locally available traffic data and the traffic condition of the neighboring intersections. According to this approach, our system consists of agents distributed into a hierarchical architecture. Each agent is responsible for one activity, for example, traffic data collection, preprocessing of these data and decision making of econfiguration of the traffic-lights controllers. An intelligent algorithm based on the policy management model is defined and used as an auxiliary element for the coordination mechanism in order to form an adaptive control system with learning capabilities that allows a more fluid traffic and reduce some of the problems that the society face such as the average wait time and trip travel time and the average size queue per intersection. We define a model based on configuration profiles that are used as initial configurations and new configuration profiles are created as the traffic conditions change. Our approach is experimented with traffic control of a few connected junctions and the result obtained is promising; it can reduce the average delayed time of each car at each traffic-light near an intersection rather substantially when compared with the current traffic-lights control approach. ; Postprint (published version)


    Zugriff

    Download


    Exportieren, teilen und zitieren



    A policy-based multi-agent management approach for intelligent traffic-light control

    Guerrero-Ibanez, A. / Contreras-Castillo, J. / Buenrostro, R. et al. | Tema Archiv | 2010


    A Policy-Based Multi-Agent Management Approach for Intelligent Traffic-Light Control, pp. 694-699

    Guerrero Ibanez, J.A. / Buenrostro Mariscal, R. / Contreras Castillo, J.J. et al. | British Library Conference Proceedings | 2010


    Intelligent traffic light control using distributed multi-agent Q learning

    Liu, Ying / Liu, Lei / Chen, Wei-Peng | IEEE | 2017


    An Approach to Intelligent Traffic Management System Using a Multi-agent System

    Hamidi, Hodjat / Kamankesh, Ali | Springer Verlag | 2017