Origin-Destination Pattern Effects on Large-Scale Mixed Traffic Control via Multi-Agent Reinforcement Learning

2025-05-19

AI Summary

To address the insufficient coordination between conventional traffic signals and connected autonomous vehicles (CAVs) in large-scale mixed vehicular-pedestrian networks, this paper proposes a decentralized multi-agent reinforcement learning (MARL) framework. Moving beyond isolated intersection modeling, the framework innovatively incorporates origin-destination (OD) flow pattern regulation into a global congestion mitigation mechanism, enabling distributed decision-making. It is rigorously evaluated in a digital twin of a real-world 14-intersection urban network in Colorado Springs. Experiments demonstrate significant reductions in average vehicle waiting time, validating both the effectiveness and engineering deployability of OD-guided cooperative control in city-scale mixed-traffic scenarios. Key contributions include: (1) an end-to-end cooperative control paradigm tailored to real-scale mixed-traffic networks; (2) a novel global optimization strategy driven by OD flow dynamics; and (3) a lightweight, scalable MARL architecture that balances algorithmic performance with system extensibility.

Technology Category

• Multiagent Systems: Multiagent Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Traffic congestion remains a major challenge for modern urban transportation, diminishing both efficiency and quality of life. While autonomous driving technologies and reinforcement learning (RL) have shown promise for improving traffic control, most prior work has focused on small-scale networks or isolated intersections. Large-scale mixed traffic control, involving both human-driven and robotic vehicles, remains underexplored. In this study, we propose a decentralized multi-agent reinforcement learning framework for managing large-scale mixed traffic networks, where intersections are controlled either by traditional traffic signals or by robotic vehicles. We evaluate our approach on a real-world network of 14 intersections in Colorado Springs, Colorado, USA, using average vehicle waiting time as the primary measure of traffic efficiency. Results demonstrate that strategically adjusting major origin-destination (OD) flow patterns can effectively reduce congestion, offering a new pathway for enhancing urban mobility.

Problem

• Large-scale mixed traffic control with human and robotic vehicles
• Decentralized multi-agent reinforcement learning for traffic management
• Optimizing origin-destination patterns to reduce urban congestion

Innovation

• Decentralized multi-agent reinforcement learning framework
• Large-scale mixed traffic network management
• Strategic OD flow pattern adjustment