Understanding and Modelling Urban Dynamics: a Human Mobility approach

Cities are complex, ever-changing systems shaped by how people move—both in their daily routines and in long-term shifts like moving homes. This thesis explores how these mobility patterns interact with digital technologies and social and economic forces, using a mix of traditional complex systems theory and modern AI techniques. We present a new AI-based model that creates synthetic but realistic data to study daily traffic flows, along with a method to assess the quality of this data. We also examine how AI-powered tools like navigation apps affect traffic patterns, finding that widespread use can unexpectedly increase CO₂ emissions. In the long term, mobility patterns feed back into broader societal changes. For example, socioeconomic conditions influence where people choose to live, and these choices, in turn, reshape those conditions. We study segregation by enhancing classic models with real-world mobility rules, showing this reduces how fast and how deeply segregation takes hold. For gentrification—the process where rising costs push out lower-income residents—we develop a new model based on simulations and time-based networks. Our findings highlight the outsized influence of the ultra-wealthy in triggering these changes and introduce new indicators to detect early signs of gentrification. By combining AI and complex systems theory, this research provides new tools and insights to understand and guide urban development in fairer and more sustainable ways.