Autonomous driving in urban settings requires intelligent decision-making ability to deal with complex behaviors in dense traffic scenarios. Traditional modular methods address these challenges using classical rule-based approaches but require heavy engineering efforts to scale to diverse and unseen environments. Recently, Deep Reinforcement Learning (DRL) has provided a data-driven framework for decision making and has been applied to urban driving. However, prior works that employ end-to-end DRL with high-dimensional sensor inputs report poor performance on complex urban driving tasks. In this work, we present a framework that combines modular and DRL approaches to solve the planning and control subproblems in urban driving. We design an input representation that enables our DRL agent to learn the complex urban driving tasks of lane-following, driving around turns and intersections, avoiding collisions with other dynamic actors, and following traffic light rules. The agent learned using our proposed approach achieves state-of-the-art performance on the NoCrash benchmark in the CARLA urban driving simulator.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Learning Urban Driving Policies using Deep Reinforcement Learning


    Contributors:


    Publication date :

    2021-09-19


    Size :

    1693142 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Autonomous Driving using Deep Reinforcement Learning in Urban Environment

    Hashim Shakil Ansari / Goutam R | BASE | 2019

    Free access

    Urban Driving with Multi-Objective Deep Reinforcement Learning

    Li, Changjian / Czarnecki, Krzysztof | ArXiv | 2018

    Free access

    Model-free Deep Reinforcement Learning for Urban Autonomous Driving

    Chen, Jianyu / Yuan, Bodi / Tomizuka, Masayoshi | IEEE | 2019


    Learning Delta Policies for Automated Driving Via Reinforcement Learning

    Templer, Maximilian / Kaste, Jonas / Hochrein, Paul et al. | TIBKAT | 2021


    DEEP REINFORCEMENT LEARNING FOR OPTIMIZING CARPOOLING POLICIES

    JINDAL ISHAN / QIN ZHIWEI / CHEN XUEWEN et al. | European Patent Office | 2019

    Free access