Skip to Main Content
This study presents a method that combines travel-time estimation and adaptive traffic signal control. The proposed method explores the concept of vehicle-to-infrastructure communication, through which real-time vehicle localisation data become available to traffic controllers. This provides opportunity to frequently sample vehicle location and speed for online travel-time estimation. The control objective is to minimise travel time for vehicles in the system. The proposed method is based on approximate dynamic programming, which allows the controller to learn from its own performance progressively. The authors use micro-traffic simulation to evaluate the control performance against benchmark control methods in an idealistic environment, where errors in sampling vehicle location and speed are not considered. The results show that the proposed method outperforms benchmarking methods substantially and consistently.