The invention provides an urban ecological road space driving behavior optimization method based on reinforcement learning. The method comprises the following steps: step 1, obtaining a simulation driving environment interacting with a Q-Learning algorithm; 2, designing a state space based on a simulation driving environment; step 3, designing an action space based on the simulated driving environment; 4, designing a reward function based on the state space; and step 5, training the Q-Learning algorithm until convergence, and updating the Q-Learning algorithm by using a greedy strategy in the training process. The method has good fitness to an actual scene, the solving efficiency of the network connection vehicle operation model is improved, the greedy decision algorithm is introduced to be combined with the reinforcement learning algorithm, the randomness of the algorithm in iterative learning is weakened, an exploration mechanism is established, the learning iteration speed of the algorithm is increased, and the network connection vehicle operation model solving efficiency is improved. And the limitation of solving the ecological road vehicle operation model caused by the animal passing problem in the prior art is made up.

    本发明提出了一种基于强化学习的城市生态道路空间驾驶行为优化方法,包括以下步骤:步骤1、获取与Q‑Learning算法进行交互的仿真驾驶环境;步骤2、基于仿真驾驶环境,设计状态空间;步骤3、基于仿真驾驶环境,设计动作空间;步骤4、基于状态空间,设计奖励函数;步骤5、训练Q‑Learning算法至收敛,在训练过程中Q‑Learning算法使用贪婪策略进行更新。本发明对实际场景有较好的贴合性,提升了网联车辆运行模型的求解效率,并引入贪婪决策算法结合强化学习算法,减弱算法在迭代学习中的随机性,建立探索机制,加快算法的学习迭代速度,弥补了现有的由于动物穿行问题造成的生态道路车辆运行模型求解的局限。


    Access

    Download


    Export, share and cite



    Title :

    Urban ecological road space driving behavior optimization method based on reinforcement learning


    Additional title:

    基于强化学习的城市生态道路空间驾驶行为优化方法


    Contributors:
    ZENG XIAOQING (author) / ZHU MINGCHANG (author) / GUO KAIYI (author) / DENG BO (author) / FENG DONGLIANG (author)

    Publication date :

    2024-11-15


    Type of media :

    Patent


    Type of material :

    Electronic Resource


    Language :

    Chinese


    Classification :

    IPC:    G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen



    Learning Urban Driving Policies using Deep Reinforcement Learning

    Agarwal, Tanmay / Arora, Hitesh / Schneider, Jeff | IEEE | 2021


    Deep reinforcement-learning-based driving policy for autonomous road vehicles

    Makantasis, Konstantinos / Kontorinaki, Maria / Nikolos, Ioannis | IET | 2019

    Free access

    Deep reinforcement‐learning‐based driving policy for autonomous road vehicles

    Makantasis, Konstantinos / Kontorinaki, Maria / Nikolos, Ioannis | Wiley | 2020

    Free access

    Deep Reinforcement-Learning-based Driving Policy for Autonomous Road Vehicles

    Makantasis, Konstantinos / Kontorinaki, Maria / Nikolos, Ioannis | ArXiv | 2019

    Free access

    Urban road network integrated control method based on coordination reinforcement learning

    FANG JIE / YOU YA / XU MENGYUN et al. | European Patent Office | 2024

    Free access