Enhancing Urban Mobility Infrastructure through Traffic Flow Modeling and Sensor-based Monitoring

Traffic simulation. This thesis deals with Traffic State Estimation in the case of large-scale urban traffic networks and aims to develop an innovative concept for the digital analysis of complex urban realities, aimed at decision support for their planning and management. The digital system combines the main physical subsystems of the city as many simulators based on artificial intelligence models, which interact with each other in the digital world, and which compose a digital twin of the city itself, continuously learning from multiple sensory sources and updating to represent the state of the physical city in real time. More specifically, this thesis is aimed at solving two main problems: how to find the optimal location of sensors, and how to incorporate heterogeneous data sources into a flow and density estimation approach. In the context of the state of the art, most TSE work only considers the case of highways, and work dealing with general networks often requires complex operations unsuited to very large systems or involves the use of very rich data that may not be available for most applications. Therefore, efficient methods for sensor localization and state estimation using commonly used data sources.