With the expansion of urban scale and the increase motor vehicles, traffic congestion is becoming more serious. The development of public transportation is the main way to alleviate congestion. As part of public transportation, taxis are popular because of their all-weather, comfort and high accessibility, but currently they mainly rely on driver experience to find passengers, resulting in imbalance between supply and demand, affecting passenger experience. In this paper, we propose a taxi dispatch model using reinforcement learning method, which utilizes the Dueling DQN algorithm to optimize the dispatch strategy to alleviate the imbalance between supply and demand. Using the GPS data of taxis in Shenzhen, after data cleaning and analysis, the ARIMA model is used to predict the future boarding and alighting volume, and a supply and demand matrix is constructed. Experiments show that the dispatch efficiency of Dueling DQN can reach 100%, and the average dispatch efficiency is 85.3%.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Urban Taxi Dispatching Optimization: A Reinforcement Learning-Based Approach


    Contributors:
    Yu, Yue (author) / Zhang, Kai (author) / Zhu, Guowei (author)


    Publication date :

    2024-12-20


    Size :

    1139391 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    DEEP REINFORCEMENT LEARNING-BASED TAXI DISPATCHING METHOD AND SYSTEM

    LIU ZHIDAN / LI JIANGZHOU / WU KAISHUN | European Patent Office | 2021

    Free access

    Context-Aware Taxi Dispatching at City-Scale Using Deep Reinforcement Learning

    Liu, Zhidan / Li, Jiangzhou / Wu, Kaishun | IEEE | 2022


    A Clustering-Based Multi-Agent Reinforcement Learning Framework for Finer-Grained Taxi Dispatching

    Rajeh, Taha M. / Luo, Zhipeng / Javed, Muhammad Hafeez et al. | IEEE | 2024



    A Smart Taxi Dispatching System

    Awang, R.-M. / ITS Congress Association | British Library Conference Proceedings | 2000