Understanding the intentions of a driver at an intersection is a useful property for a variety of driver assistance functions ranging from engine management to safety. This paper shows how a time series prediction can be modelled to classify the intended stop length with high accuracy in a realworld driving situation. Stop length is defined by how many seconds a vehicle waits at a position before continuing its journey. An LSTM was activated when the vehicle comes to a stop and processes the last few seconds of data signals from the vehicle. Many possible input signals have been explored. It was shown that specifically the time series of the brake pressure and the steering wheel angle contribute the most to an accurate prediction.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    Driver intention prediction of stop length using LSTM


    Beteiligte:


    Erscheinungsdatum :

    19.09.2021


    Format / Umfang :

    439476 byte




    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Real-Time Driver Maneuver Prediction Using LSTM

    Khairdoost, Nima / Shirpour, Mohsen / Bauer, Michael A. et al. | IEEE | 2020


    Driver intention prediction and intention trajectory updating method

    LI JUNQIU / WU YONGHUA / CHEN JIAHUI | Europäisches Patentamt | 2023

    Freier Zugriff

    Prediction of driver's intention to stop for engine start/stop

    TANG CHUANCHI / THOMBARE PAPEEHA / MAYALANKAR DRUSHAN et al. | Europäisches Patentamt | 2023

    Freier Zugriff

    STOP INTENTION DETERMINATION APPARATUS AND STOP INTENTION DETERMINATION METHOD

    AOI HATSUMI / OKAJI KAZUYOSHI / SUGAHARA HIROSHI et al. | Europäisches Patentamt | 2018

    Freier Zugriff

    STOP INTENTION DETERMINATION DEVICE AND STOP INTENTION DETERMINATION METHOD

    AOI HATSUMI / OKAJI KAZUYOSHI / SUGAWARA HIROSHI et al. | Europäisches Patentamt | 2018

    Freier Zugriff