Integrating independent and centralized multi-agent reinforcement learning for traffic signal network optimization

TL;DR

This paper introduces QCOMBO, a multi-agent reinforcement learning algorithm that combines independent and centralized learning to optimize traffic signals across large networks, demonstrating scalability and generalization in traffic control.

Contribution

The paper proposes QCOMBO, a novel MARL algorithm that improves scalability and global performance in traffic signal optimization by integrating independent and centralized learning with a consistency regularization.

Findings

QCOMBO outperforms recent MARL algorithms in diverse traffic scenarios.

Policies trained on small networks generalize well to larger networks.

QCOMBO demonstrates effective traffic flow improvements in SUMO simulations.

Abstract

Traffic congestion in metropolitan areas is a world-wide problem that can be ameliorated by traffic lights that respond dynamically to real-time conditions. Recent studies applying deep reinforcement learning (RL) to optimize single traffic lights have shown significant improvement over conventional control. However, optimization of global traffic condition over a large road network fundamentally is a cooperative multi-agent control problem, for which single-agent RL is not suitable due to environment non-stationarity and infeasibility of optimizing over an exponential joint-action space. Motivated by these challenges, we propose QCOMBO, a simple yet effective multi-agent reinforcement learning (MARL) algorithm that combines the advantages of independent and centralized learning. We ensure scalability by selecting actions from individually optimized utility functions, which are shaped to maximize global performance via a novel consistency regularization loss between individual utility and a global action-value function. Experiments on diverse road topologies and traffic flow conditions in the SUMO traffic simulator show competitive performance of QCOMBO versus recent state-of-the-art MARL algorithms. We further show that policies trained on small sub-networks can effectively generalize to larger networks under different traffic flow conditions, providing empirical evidence for the suitability of MARL for intelligent traffic control.