Research

Transportation AI

I build scalable models and decision systems for transportation systems where demand, control, energy use, and security risks interact. A central thread in my work is using AI to design operational policies that remain practical at network scale.

Representative work: Scalable reinforcement learning approaches for dynamic pricing in ride-hailing systems, Transportation Research Part B. Project: umnilab/RL_pricing.

Network-Level Simulation

I develop simulation platforms and simulation-supported reasoning methods for evaluating emerging mobility systems before they are deployed in real cities.

METS-R SIM is an open-source city-scale simulation platform for Multi-modal Energy-optimal Trip Scheduling in Real-time. The platform supports shared autonomous electric vehicles, taxis, buses, charging operations, routing, and service evaluation.

This work grew from the U.S. Department of Energy METS-R project, where I served as student lead and major contributor. I use the platform to evaluate energy, congestion, and service impacts of operational algorithms for electric mobility services and transportation hubs.

Representative work: METS-R SIM: A simulator for multi-modal Energy-optimal Trip Scheduling in Real-time with shared autonomous electric vehicles, Simulation Modelling Practice and Theory. Project documentation: METS-R docs.

CPS Operation and Security

I study how socio-cyber-physical transportation systems behave under strategic, rare, and adversarial conditions. The goal is to make mobility services efficient while also understanding operational and security risks before they become system-level failures.

My Ph.D. research studied how mobility-on-demand systems can operate efficiently while remaining robust to strategic and adversarial behavior. This work includes scalable reinforcement learning approaches for dynamic pricing, data-driven fleet relocation, and network-level analysis of adversarial booking attacks against autonomous mobility-on-demand services.

Representative work: Assessing the risks of adversarial booking attack to autonomous mobility-on-demand services, Transportation Research Part C. Project: umnilab/ADV_Booking.

Mobility Analytics for Disasters and Recovery

I analyze large-scale GPS and trajectory data to understand how travelers respond to disruptions and compound risks. Recent work models evacuation and return behavior during Hurricane Ida under COVID-19 conditions, transportation demand recovery, hub operations, and resilience in urban mobility systems.

Current Directions

• Simulation-based verification with METS-R SIM, CARLA, and autonomous driving stacks.
• Scenario search and amortized sampling for rare, high-impact transportation events.
• Cybersecurity risks in next-generation road transportation.
• AI methods for transportation control under uncertainty.