Scholarlink

In public transportation, bus lanes, as an important means of transportation and facilities, can reduce urban traffic congestion and improve the service quality and operation efficiency of public transportation while ensuring high speed, efficiency and high quality. Aiming at the demand of coordination and optimization of signal control between intermittent bus lanes and downstream signalized intersections, this paper proposes a Multi-Type Multi-Agent hybrid control model covering bus lane Agent, bus Agent and traffic signal light Agent. Two reinforcement learning methods, Q learning and SARSA, are used to design parameters for the two control scenarios of traditional intermittent bus lanes and mobile intermittent bus lanes, and the optimization models of TLS-IBL-QL, TLS-IBL-SARSA, TLS-MBL-QL and TLS-MBL-SARSA are proposed. The experimental results show that the performance of TLS-IBL-SARSA algorithm is better than that of TLS-IBL-QL algorithm, and that of TLS-MBL-QL is better than that of TLS-MBL-SARSA algorithm. Both kinds of control algorithms are obviously better than the traditional TLS-DBL bus lane control.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Coordinated Control Method for Intermittent Bus Lanes Based on Reinforcement Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Guorong Zheng,
  • Qingwan Xue,
  • Junjie Fu,
  • Han Liu

摘要

In public transportation, bus lanes, as an important means of transportation and facilities, can reduce urban traffic congestion and improve the service quality and operation efficiency of public transportation while ensuring high speed, efficiency and high quality. Aiming at the demand of coordination and optimization of signal control between intermittent bus lanes and downstream signalized intersections, this paper proposes a Multi-Type Multi-Agent hybrid control model covering bus lane Agent, bus Agent and traffic signal light Agent. Two reinforcement learning methods, Q learning and SARSA, are used to design parameters for the two control scenarios of traditional intermittent bus lanes and mobile intermittent bus lanes, and the optimization models of TLS-IBL-QL, TLS-IBL-SARSA, TLS-MBL-QL and TLS-MBL-SARSA are proposed. The experimental results show that the performance of TLS-IBL-SARSA algorithm is better than that of TLS-IBL-QL algorithm, and that of TLS-MBL-QL is better than that of TLS-MBL-SARSA algorithm. Both kinds of control algorithms are obviously better than the traditional TLS-DBL bus lane control.