Urban mobility foundation model: A literature review and hierarchical perspective

Abstract

An urban mobility system serves as a highly intricate and nonlinear mega-system facilitating the movement of people, goods, and services across spatio-temporal domains. This complexity stems from factors such as intricate interactions between transportation supply and demand, and the inherent stochastic nature of an open, heterogeneous, and adaptable system. Successfully comprehending and navigating this system presents a challenge. Yet, a remarkable opportunity emerges with the growing availability of multi-source data in urban mobility and various sectors, combined with the recent advancements in large-scale machine learning (ML) models. In this paper, we introduce a novel conceptual framework, the HUGE (Hierarchically Unified GEnerative) foundation model, to address multifaceted computational tasks and decision-making problems embedded in urban mobility systems. We delve into the core technologies and their seamless integration to realize this framework, highlighting its potential to harness substantial data analytics, hierarchical ML methodologies, and domain-specific knowledge. The conceived framework has the potential to revolutionize urban mobility system planning, design, construction, and management in a digital and intelligent manner.