Urban AI & Digital Twin Research
About
Our work brings together data-driven discovery, intelligent modeling, and system-level optimization to understand and shape the movement of people and goods in modern cities. By uniting advances in artificial intelligence, large-scale simulation, and network analytics, we create tools that can capture the complexity of real-world mobility, anticipate how it will evolve, and test how different interventions might perform before they are implemented. This integrated perspective allows us to move beyond isolated analyses—linking individual behavior, collective dynamics, and infrastructure performance into a single, adaptive view of the transportation ecosystem.
The ultimate goal is to create mobility systems that are smarter, more adaptive, and more intelligent—systems that not only respond to real-time conditions but also anticipate future needs and challenges. By leveraging AI, data fusion, and simulation at scale, our work provides powerful tools for planners, policymakers, and communities to design transportation networks that reduce congestion, lower emissions, improve safety, and expand access for all. From supporting clean energy transitions and optimizing infrastructure investments to ensuring underserved populations are included in mobility planning, our research strives to align technological advancement with public benefit—helping cities and regions build sustainable, efficient, and inclusive transportation systems for the future.
AOI 1 – Mobility Network Optimization
We design smart, AI-powered systems that keep our transportation networks running smoothly—no matter what challenges arise. From planning where electric vehicle chargers should go, to orchestrating large-scale recovery efforts after natural disasters, to simulating the complex movement of millions of people and goods in a megacity, our research tackles the intricate puzzle of modern mobility. Using cutting-edge simulation frameworks, we model entire urban regions—down to individual travel decisions—allowing us to test and refine strategies before they’re applied in the real world. By blending advanced simulations, real-time data, and intelligent decision-making, we create mobility networks that can adapt to demand changes, recover quickly from disruptions, and serve communities equitably. The ultimate goal: transportation systems that are faster, safer, more resilient, and ready for a sustainable future. Below are some projects that we have done in the area:
- Multi-Agent Multimodal Transportation Simulation for Mega-cities: Application of Los Angeles
- Multi-Period Truck Scheduling with Queueing for Post-disaster Debris Removal: A Case Study of the 2025 Los Angeles Wildfires (ITSC 2025)
- Large-Scale Public Charging Demand Prediction with a Scenario-and Activity-Based Approach
AOI 2 – AI-Empowered Spatiotemporal Mobility Intelligence
Understanding how, when, and why people move through cities is essential for designing transportation systems that are efficient, equitable, and resilient. We aim to transform how cities understand and anticipate human movement by uniting advanced mobility data mining with generative AI. Our work integrates heterogeneous sources—such as GPS traces, travel surveys, points of interest, demographic data, and infrastructure networks—into cohesive, semantically rich representations of travel behavior. By combining deep learning, large language models, and transferable generative frameworks, we can extract patterns from massive datasets, reconstruct incomplete or sparse information, and simulate realistic mobility scenarios at fine spatial and temporal resolutions. These methods capture both individual and collective decision-making, account for diverse socio-demographic contexts, and adapt seamlessly to regions with limited data. The result is a flexible, privacy-preserving foundation for analyzing current travel behavior, forecasting future demand, and supporting equitable, data-informed transportation planning at metropolitan and global scales. Some of our projects and papers in the area are listed below:
- Deep Activity Model: A Generative Approach for Human Mobility Pattern Synthesis (Under Review)
- Semantic Trajectory Data Mining with LLM-Informed POI Classification (ITSC 2024 Best Paper)
- Learning Universal Human Mobility Patterns with A Foundational Model for Cross-Domain Data Fusion (Transportation Part C Special Issue: Foundational Models and Large Language Model for Human Mobility)
- Next-Generation Travel Demand Modeling with a Generative Framework for Household Activity Coordination (ITSC 2025)
AOI 3 – Urban Mobility Digital Twins
Construct high-fidelity city digital twins, including people, vehicles, and infrastructure, providing a risk-free sandbox on a large scale for AI training, scenario exploration, and evaluation for policy and strategy. One of our projects in the area is:
AOI 4 – AI Agents & Network Mobility Intelligence
This research area focuses on creating intelligent, interactive digital representations of travelers, vehicles, and transportation networks to better understand, predict, and optimize urban mobility. By combining advanced AI agent modeling with adaptive network intelligence, we simulate realistic decision-making, coordinated activities, and multimodal travel behaviors at scale. These systems learn from diverse data sources—ranging from surveys and GPS traces to infrastructure and traffic records—and operate within dynamic, data-driven digital twins of mobility systems. Through continuous interaction between agents and networks, the models can adapt to evolving conditions, test policy or infrastructure scenarios, and support inclusive, data-informed planning. This integrated approach enables not only more accurate mobility forecasting, but also the development of responsive, equitable, and efficient transportation strategies for complex, real-world environments. Listed below are some of our works in the area:
- Mobility AI Agents and Networks (IEEE Transactions on Intelligent Vehicles)
- Human Mobility Modeling with Household Coordination Activities under Limited Information via Retrieval-Augmented LLMs (ITSC 2025)
- Beyond 9-to-5: A Generative Model for Augmenting Mobility Data of Underrepresented Shift Workers (ITSC 2025)