Self-Scheduling Agents for Real-Time Traffic Signal Control

Abstract

Optimizing traffic signal control in real-time to respond to changing vehicle flows is a challenging problem. As the size of the road network increases, the exponential growth in joint signal timing and traffic states presents a difficult computational barrier to determining effective traffic light signalization. Furthermore, the dynamic nature of traffic flows makes reliable prediction possible only over a limited time horizon, and hence necessitates continual re-computation and adaptation of computed solutions. In this paper, we take a self-scheduling approach to solving the traffic signal control problem, where each intersection is controlled by a self-interested agent operating with a limited (fixed horizon) view of incoming traffic. Central to the approach is a representation that aggregates incoming vehicles into critical clusters, based on the non-uniform distributed nature of road traffic flows. Such aggregation of vehicles into anticipated queues and platoons enables the real-time computation of signal timing policies that incorporate greater look-ahead than traditional queue clearing strategies and better promote the establishment of “green waves” where vehicles flow through the road network without stopping. We present simulation results on some dynamic traffic scenarios that demonstrate the leverage that our approach provides over two state-of-the-art self-organizing approaches.