Artificial Intelligence in Smart Traffic Management: A Systematic Review

This paper presents a comprehensive review of Artificial Intelligence (AI) applications in modern traffic management systems. As global urbanization accelerates, traditional traffic control infrastructures are increasingly overwhelmed by the demands of high-density vehicular flow, resulting in congestion, pollution, and road safety challenges. AI-driven frameworks encompassing deep reinforcement learning, computer vision, graph neural networks, and predictive analytics are rapidly transforming how cities manage traffic at scale. This review examines the theoretical underpinnings, core architectural components, and real-world deployments of AI-based traffic systems. We analyze intelligent signal control, vehicle detection, incident prediction, route optimization, and connected vehicle integration as primary domains of AI application. Additionally, we survey three operational paradigms: Centralized AI Control, Distributed Edge Intelligence, and Hybrid Cloud-Edge Systems, evaluating their respective trade-offs in latency, scalability, and resilience. The paper provides an analysis of comparative architectures alongside discussion of open research challenges and future directions in autonomous urban mobility.