V3I6P111

Artificial Intelligence–Enabled Traffic Optimization in Smart Cities: An Integrative Scientific Review, Evidence Synthesis, and Conceptual Framework for Sustainable Urban Mobility

Eyad Al Haddad^1*

Abstract

The high rate of urbanization, motorization, and more complex commuter patterns have increased congestion, air pollution, and safety hazards of metropolitan road networks. Traditional traffic management systems are still limited by a poor sensory coverage, non-adaptive or weakly adaptive control software, and decentralized operations in between intersections and corridors. Artificial Intelligence (AI) is also becoming a core technology of the next-generation intelligent transportation systems (ITS) due to its ability to contextualize volatile, high-volume, and heterogeneous streams of data, which facilitates the continuous estimation of traffic states, prediction, and adaptive control. The paper is an integrative scientific review of AI-based traffic optimization in smart cities with a particular focus on (i) spatiotemporal traffic prediction of machine learning, deep learning and graph, (ii) adaptive traffic signal control of reinforcement learning and multi-agent coordination (iii) computer vision of real-time perception, incident detection, and compliance monitoring. The documented deployments are employed to put the achievable impacts into perspective; an example of this is the Los Angeles ATSAC program that has had travel times reported to decrease by the order of 10 percent in certain evaluations and the number of stops and delay cut in the previous performance reports. It suggests a conceptual framework that incorporates edge perception, centralized and federated analytics, predictive decision support, and adaptive control together with sound governance and evaluation measures. The synthesis shows that AI-based traffic systems could provide quantifiable efficiency and safety improvements, yet their performance relies on the quality of data, model generalization, human-in-the-loop control, cybersecurity, and maintenance of the lifecycle. The paper ends in a set of research directions of reliable, scalable and sustainable AI in urban mobility.