Luo, Rui (Beijing jiaotong university), ShangGuan, Wei (Beijing Jiaotong University), Chai, Linguo (Beijing Jiaotong University), Chen, Junjie (Beijing Jiaotong University)

Cooperative Control for Signals and Vehicles Based on Deep Reinforcement Learning in Mixed Traffic Environments

Scheduled for presentation during the Regular Session "S36a-Behavior Modeling and Decision-Making in Traffic Systems" (FR-LM-T36), Friday, November 21, 2025, 11:30−11:50, Surfers Paradise 3

2025 IEEE 28th International Conference on Intelligent Transportation Systems (ITSC), November 18-21, 2025, Gold Coast, Australia

This information is tentative and subject to change. Compiled on October 18, 2025

Abstract

The traffic control system plays an import role on reducing traffic conflicts and improving traffic efficiency at intersections. With the development of vehicle-Infrastructure-cloud integrated system, collaboration between signals and vehicles has the potential to further improve the efficiency of intersections. This paper proposed a cooperative control method for signals and vehicles (CCSV) based on deep reinforcement learning in mixed traffic environments. This method includes a signal-agent and a vehicle controller. The signal-agent will obtain the traffic flow rate, penetration rate in the intersection, and vehicle information to decide the optimal signal timing schedule. The selected phases and phase duration are dynamically adjusted based on current traffic conditions, which can achieve more flexible signal control. The vehicle controller will generate the CAV eco-driving trajectories according to the optimal signal timing. Simulations were conducted in a mixed traffic environment with varying traffic demand and penetration rate. The simulation results show that our method can reduce the fuel consumption by up to 29.0%, waiting time by up to 56.6% and throughput by up to 2.6%.