DTUMOS, Digital Twin for Large-scale Urban Mobility Operating System

Abstract The rapid development of digital twin technology has significantly changed the way virtual cities are used in smart cities and the transportation. In particular, digital twins provide a playground where various mobility systems, algorithms, and policies can be developed and tested. In this study, DTUMOS, a digital twin framework for urban mobility operating systems, is proposed. We construct open-source framework that can easily and flexibly apply to any city and mobility system worldwide. A novel architecture that combines an AI-based estimated time of arrival model and vehicle router algorithm enables DTUMOS to achieve high-speed performance while maintaining accuracy when implementing large-scale mobility systems. The proposed DTUMOS has distinct strengths in scalability, speed, and visualization compared to the existing state-of-the-art mobility digital twins. The performance and scalability are verified by using actual data in large metropolitan cities, such as Seoul, New York City, and Chicago. A lightweight and open-source environment of DTUMOS opens a new era for developing various simulation-based algorithms and quantitatively evaluating the effectiveness of policies for future mobility systems.