Research

Models and algorithms for adaptive mobility systems.

The Tsinghua Mobility Science (MoS) Lab studies how people, infrastructure, and policy interact across transportation systems, especially under uncertainty, disruption, and rapid urban change.

Featured Research

Schematic of the individual path recommendation problem on a disrupted rail and bus network (Figure 1, Transportation Science)
Public Transit Resilience

Individual path recommendation under transit disruptions

Models passenger behavior uncertainty and recommends resilient paths during service disruptions.

Commuting patterns in Beijing, Munich, and Singapore before and after housing exchange (Figure 2, Nature Sustainability)
Sustainable Urban Systems

Housing exchange to reduce commuting emissions

Uses household-level housing exchange strategies to reduce excess commuting emissions.

Pair-wise attention-based pointer neural network architecture (Figure 2, Transportation Research Part E)
Mobility AI

Last-mile delivery route prediction

A pair-wise attention-based pointer neural network that predicts drivers' route trajectories in last-mile delivery.

Physical network and time-space hypernetwork representation (Figure 1, Transportation Science)
Public Transit Resilience

Ex post path choice estimation

Estimates urban rail passenger path choices from smart card data via an aggregated time-space hypernetwork.

Research Areas

01

Public Transit Resilience

We study how transit systems and passengers respond to planned and unplanned disruptions. Our work includes incident-aware passenger behavior inference, robust path recommendation, network performance modeling, and control strategies that reduce congestion and improve service reliability.

02

Travel Behavior & Demand

We build interpretable and predictive models of mobility behavior using smart card data, license plate recognition data, surveys, and operational data. Topics include route choice, mode choice, individual mobility prediction, and the interactions between public transit and emerging mobility services.

03

Mobility AI

We apply machine learning, deep learning, robust optimization, and decision models to transportation problems such as estimated time of arrival, last-mile delivery route prediction, time-series forecasting, and classification under data uncertainty.

04

Sustainable Urban Systems

We analyze transportation as part of broader cyber-physical-social urban systems, with applications in commuting emissions, extreme weather resilience, public health risk during commuting, housing mobility, and policy evaluation.