Abstract Traffic light control is an issue that metropolitan cities are confronted with because of ever-increasing vehicles growth. A proper timing strategy can release the congestion level and minimize vehicle delays better than standard fixed signal timing (FST) strategy. In this paper, a self-adaptive Q-learning based traffic light green time control strategy (QTGCS) is designed to reduce the congestion. Q-Learning acts as the learning mechanism for light at each intersection as an independent agent to release itself by operating traffic light duration dynamically. Q-Learning is based on discrete representation of state and action spaces. In order to make QTGCS not independent of the environment and designer criterion, Fuzzy Q-learning based traffic light green time control strategy (FQTGCS) is proposed by applying the fuzzy logic control to optimize timing strategy based on the number of vehicles receiving from the adjacent intersections. We show through simulation that our algorithm effectively over the real road networks.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Traffic Lights Dynamic Timing Algorithm Based on Reinforcement Learning


    Contributors:
    Lu, Chenqing (author) / Wen, Feng (author) / Gen, Mitsuo (author)


    Publication date :

    2017-06-29


    Size :

    10 pages




    Type of media :

    Article/Chapter (Book)


    Type of material :

    Electronic Resource


    Language :

    English




    Timing traffic lights correctly

    Neill, C.W. | Engineering Index Backfile | 1929


    Dynamic timing method and device for traffic lights

    WU JIANBO / KU YABO / ZHAO JIA et al. | European Patent Office | 2023

    Free access

    Signal Synchronization of Traffic Lights Using Reinforcement Learning

    Aydin, Ilhan / Sevi, Mehmet / Gungoren, Gurbet et al. | IEEE | 2022


    Traffic Lights Control Using Reinforcement Learning: A Comparative Study

    Errajraji, Khalid / Bouayad, Anas / Fardousse, Khalid | Springer Verlag | 2024