Scalable Adaptive Traffic Light Control Over a Traffic Network Including Turns, Transit Delays, and Blocking

TL;DR

This work tackles multi-intersection traffic light control on a grid, explicitly modeling turning movements, transit delays between intersections, and blocking propagation. It introduces a stochastic hybrid system model and derives Infinitesimal Perturbation Analysis (IPA) gradient estimators of the cost with respect to controllable parameters $\Theta = \{\theta_p^{min}, \theta_p^{max}, s_p\}$, enabling online gradient-based updates. The main contributions are a flexible, scalable modeling framework that integrates turns, delays, and blocking, and an online controller that adapts in real time using only event-based observations; SUMO-based simulations show improvements in mean waiting times and synchronization relative to fixed-cycle baselines, with linear scalability in network size. The approach provides a practical path toward distributed, real-time TLC in large urban networks, with potential extensions to bicycles and pedestrians and adaptive step-size tuning for faster convergence.

Abstract

We develop adaptive data-driven traffic light controllers for a grid-like traffic network considering straight, left-turn, and right-turn traffic flows. The analysis incorporates transit delays and blocking effects on vehicle movements between neighboring intersections. Using a stochastic hybrid system model with parametric traffic light controllers, we use Infinitesimal Perturbation Analysis (IPA) to derive a data-driven cost gradient estimator with respect to controllable parameters. We then iteratively adjust them through an online gradient-based algorithm to improve performance metrics. By integrating a flexible modeling framework to represent diverse intersection and traffic network configurations with event-driven IPA-based adaptive controllers, we develop a general scalable, adaptive framework for real-time traffic light control in multi-intersection traffic networks.

Figures (10)

• Figure 1: Grid traffic system
• Figure 2: Traffic Structure Demonstration
• Figure 3: State space representation
• Figure 4: Example for trajectories of $y_{q,q^d,m}(t)$ and $\Delta_{q,q^d}(t)$
• Figure 5: Typical sample path of a traffic queue