Ethical decision-making in autonomous vehicles has been a significant area of research since the emergence of the Trolley Problem. However, current studies fail to effectively incorporate the operative state of the vehicle and instead rely exclusively on sociological attributes for decision-making. This paper establishes three ethical traffic scenarios that reflect the most typical ethical dilemmas. Based on this, we examine the ethical decision-making of autonomous vehicles in each scenario. Firstly, to enable the decision-making system of autonomous vehicles to solve ethical dilemmas, a coupled ethical reward function model is innovatively proposed based on human feedback that integrates knowledge from sociology, economics, and vehicle dynamics. Furthermore, an ethics-driven multi-modal network model is proposed to extract morphological features and dynamic features from perceptual information and road test data, respectively. Finally, an ethical simulation experiment is conducted, which demonstrates that the decision-making strategies generated by the proposed model in the ethical traffic scenario are more aligned with human intentions compared to those of the control group.


    Zugriff

    Zugriff prüfen

    Verfügbarkeit in meiner Bibliothek prüfen

    Bestellung bei Subito €


    Exportieren, teilen und zitieren



    Titel :

    A Human Feedback-Driven Decision-Making Method Based on Multi-Modal Deep Reinforcement Learning in Ethical Dilemma Traffic Scenarios


    Beteiligte:
    Gao, Xin (Autor:in) / Luan, Tian (Autor:in) / Li, Xueyuan (Autor:in) / Liu, Qi (Autor:in) / Meng, Xiaoqiang (Autor:in) / Li, Zirui (Autor:in)


    Erscheinungsdatum :

    24.09.2023


    Format / Umfang :

    2292819 byte





    Medientyp :

    Aufsatz (Konferenz)


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Deep reinforcement learning-based ethical-driven multi-modal decision-making method

    LI XUEYUAN / GAO XIN / LUAN TIAN et al. | Europäisches Patentamt | 2023

    Freier Zugriff


    Cascade deep reinforcement learning security decision-making method based on multi-modal space-time representation

    YANG YUXIANG / GE FENGLONG / ZHAO JUFENG et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Intersection decision-making method based on multi-agent deep reinforcement learning

    DU YU / JIANG ANNI / ZHAO SHIXIN et al. | Europäisches Patentamt | 2024

    Freier Zugriff

    Reinforcement learning multi-lane driving decision-making method for dynamic traffic environment

    WU LIANGGUANG / ZHOU KAILONG / FEI TING | Europäisches Patentamt | 2024

    Freier Zugriff