HighlightsWe constructed new dilemma zone drivers’ decision prediction/classification models.We used two driving simulator studies datasets in our analysis.We investigated the prediction accuracy using two time frames: the last few seconds of green between 6s of travel and the onset of yellow, and during the first 3s of yellow.We found that the model using data collected during the first 3s of yellow was more accurate than the other models.

    AbstractAccurate modeling of driver decisions in dilemma zones (DZ), where drivers are not sure whether to stop or go at the onset of yellow, can be used to increase safety at signalized intersections. This study utilized data obtained from two different driving simulator studies (VT-SCORES and NADS datasets) to investigate the possibility of developing accurate driver-decision prediction/classification models in DZ. Canonical discriminant analysis was used to construct the prediction models, and two timeframes were considered. The first timeframe used data collected during green immediately before the onset of yellow, and the second timeframe used data collected during the first three seconds after the onset of yellow. Signal protection algorithms could use the results of the prediction model during the first timeframe to decide the best time for ending the green signal, and could use the results of the prediction model during the first three seconds of yellow to extend the clearance interval. It was found that the discriminant model using data collected during the first three seconds of yellow was the most accurate, at 99% accuracy. It was also found that data collection should focus on variables that are related to speed, acceleration, time, and distance to intersection, as opposed to secondary variables, such as pavement conditions, since secondary variables did not significantly change the accuracy of the prediction models. The results reveal a promising possibility for incorporating the developed models in traffic-signal controllers to improve DZ-protection strategies.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Assessment of driver stopping prediction models before and after the onset of yellow using two driving simulator datasets


    Beteiligte:

    Erschienen in:

    Erscheinungsdatum :

    2015-04-28


    Format / Umfang :

    8 pages




    Medientyp :

    Aufsatz (Zeitschrift)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch





    Study on Driver-Intent Estimation at Yellow Traffic Signal by Using Driving Simulator

    Mabuchi, R. / Yamada, K. / Institute of Electrical and Electronics Engineers | British Library Conference Proceedings | 2011


    Construction of Personalized Driver Models Based on LSTM Using Driving Simulator

    Hamada, Ayaka / Oikawa, Shoko / Hirose, Toshiya | British Library Conference Proceedings | 2022


    Construction of Personalized Driver Models Based on LSTM Using Driving Simulator

    Hamada, Ayaka / Oikawa, Shoko / Hirose, Toshiya | SAE Technical Papers | 2022