Multi-Agent Deep Reinforcement Learning for Adaptive Traffic Signal Control in Smart Cities

Smart cities are increasingly reliant on sophisticated traffic management systems to alleviate congestion and improve urban mobility. This study addresses the challenge by introducing a multi-agent deep reinforcement learning (MARL) framework for adaptive traffic signal control. Each intersection within the smart city ecosystem is conceptualized as an autonomous agent equipped with real-time sensor data inputs. These agents employ a decentralized learning strategy to refine signal timing policies, promoting efficient traffic flow. The methodology capitalizes on a comprehensive traffic dataset enriched with vehicle counts, wait times, queue lengths, and emergency vehicle data, facilitating robust training and evaluation processes. Experimental results demonstrate substantial reductions in average wait time, queue lengths, and enhancements in emergency vehicle passage time, underscoring the framework's efficacy. The principal contribution of this research is the development of an advanced MARL system that promises to transform traffic management, bolster urban mobility, and pave the way for smarter cities.