I study urban traffic congestion as a complex, data-rich transportation problem. My work combines deep learning, classical machine learning, simulation, and network-based analysis to improve how congestion is predicted, interpreted, and mitigated.

Societal Relevance and SDGs

My research contributes most directly to SDG 11, Sustainable Cities and Communities. I study urban traffic congestion because it affects the reliability, accessibility, and livability of cities. My work focuses on predicting congestion, explaining where it forms, and evaluating how future urban infrastructures can better absorb traffic-related disruptions.

The work also relates to SDG 3, Good Health and Well-being, because congestion is linked to air quality, productive time loss, and urban livability. I do not model health outcomes directly; rather, I address congestion as a transport-system factor that influences the conditions under which people live and move in cities.

A methodological part of my research contributes to SDG 12, Responsible Consumption and Production. In my work on complexity-aware traffic prediction, I study whether deep learning models can be selected according to the intrinsic complexity of the prediction task, instead of defaulting to larger models. This supports more computationally efficient and deployable traffic prediction workflows.

My work also supports SDG 17, Partnerships for the Goals, through reproducibility, open datasets, open-source code, and transferable methods. Since my 2021 survey on deep learning for traffic congestion, I have tried to use freely available datasets and release paper-specific code whenever possible, so that research can be reproduced, compared, and extended across cities.

Future Directions

Going forward, I want to continue working on transport research that is useful beyond isolated case studies. This includes developing congestion prediction methods that are transferable across cities, interpretable enough to support planning decisions, and computationally efficient enough to be deployed in settings with limited resources.

I am especially interested in three directions. First, I want to study how open and crowdsourced mobility datasets can support congestion prediction in cities where high-quality proprietary traffic data are not available. Second, I want to continue developing interpretable methods that connect model predictions to transport-system features, especially road network structure and infrastructure constraints. Third, I want to study the resilience of interconnected urban infrastructures, such as transport and electric-vehicle charging systems, under non-recurrent congestion and other disruptions.

Global Collaborators

I collaborate across institutions and disciplines on transport AI, simulation, mobility data, and infrastructure resilience.