Seongjin Choi @UMN

Two undergraduate researchers complete UROP projects: TMC-Agent and HighwayVLM

2026-04-23T00:00:00+00:00

Two talented undergraduate students worked with the lab through the Undergraduate Research Opportunities Program (UROP) this semester.

Leo Curtis worked on TMC-Agent, an agentic AI system that digests multi-source data streams — traffic detector data, aggregated probe-vehicle speeds, and camera video streams — to deliver traffic operators actionable recommendations and real-time status reports.

Ismail Yusuf worked on HighwayVLM, a highway incident-detection system that combines classical computer vision with Vision-Language Models (VLMs) in a hybrid architecture for robust, context-aware detection.

Both projects push the boundaries of how AI can support real-world transportation operations, and it’s been genuinely rewarding to see their curiosity, rigor, and ownership throughout the semester.

View the announcement on LinkedIn ↗

Featured in the CTS webinar: Preparing for AI use in transportation

2026-03-24T00:00:00+00:00

The University of Minnesota Center for Transportation Studies (CTS) hosted a webinar, Preparing for AI use in transportation, on how artificial intelligence is increasingly shaping the way transportation systems are designed, planned, and managed in Minnesota — from predicting traffic patterns to analyzing vehicle movements.

Prof. Choi was featured alongside Qizhi He (University of Minnesota) and Melissa Barnes (AI program manager, Minnesota Department of Transportation), sharing how AI currently supports transportation work and how professionals can prepare for its growing influence. Prof. Choi emphasized that AI literacy is a core skill for emerging engineers, and that the department has begun integrating AI into its curriculum as public and private agencies pilot AI for operations, maintenance, and planning.

Read the CTS article ↗ · View on LinkedIn ↗

[260112] Transportation Research Board Annual Meeting 2026

2026-01-12T00:00:00+00:00

Our lab presented 8 papers at the 105th Transportation Research Board Annual Meeting (TRB 2026) in Washington, D.C. Our research spans applications of Generative AI, large language models, and optimization for transportation sensing, estimation, and operations.

[250425] Invited talk at North Carolina Agricultural and Technical State University

2025-04-25T00:00:00+00:00

It was such a pleasure to present my works on "Generative AI for Transportation Operations" at the Department of Civil, Architectural, and Environmental Engineering, NC A&T University. I would like to thank Prof. Venktesh Pandey for the invitation.

[250424] Invited talk at North Carolina State University

2025-04-24T00:00:00+00:00

It was such a pleasure to present my works on "Generative AI for Transportation Operations" at the Department of Civil, Construction, and Environmental Engineering, NC State University. I would like to thank Prof. Ali Hajbabaie and Prof. Danjue Chen for the invitation.

[250420] New survey paper on Deep Generative Models in Transportation Research published in Transportation Research Part C

2025-04-20T00:00:00+00:00

🚀 It’s official—our new paper is live!

I’m thrilled to share that our latest publication, “𝐀 𝐆𝐞𝐧𝐭𝐥𝐞 𝐈𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧 𝐚𝐧𝐝 𝐓𝐮𝐭𝐨𝐫𝐢𝐚𝐥 𝐨𝐧 𝐃𝐞𝐞𝐩 𝐆𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐌𝐨𝐝𝐞𝐥𝐬 𝐢𝐧 𝐓𝐫𝐚𝐧𝐬𝐩𝐨𝐫𝐭𝐚𝐭𝐢𝐨𝐧 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡”, has just been published in Transportation Research Part C. This work is a true team effort from experts around the world, including Zhixiong Jin, Seung Woo Ham, Jiwon Kim, Lijun Sun, and me. We hope this paper serves as a go-to primer on Generative AI for transportation researchers.

In this paper, we provide an accessible overview of Deep Generative Models (DGMs) and their applications for transportation research communities. Our paper offers a comprehensive introduction to the foundational concepts of DGMs and a review of current literature on applications of DGMs in transportation research.

To further support the transportation research community, we have included open-source tutorial code to help researchers apply DGMs to a variety of transportation problems, featuring two widely studied problems to assist a broad range of researchers:

Generating Household Travel Survey data using the National Household Travel Survey in Korea, and
Generating highway traffic speed contour using I-24 MOTION data

🔗 Free access until July 4, 2025:

https://www.sciencedirect.com/science/article/pii/S0968090X25001494?dgcid=author

💻 Tutorial Code repo:

https://github.com/UMN-Choi-Lab/DGMinTransportation

[250307] New publication in Transportation Science

2025-03-07T00:00:00+00:00

I am pleased to share that my paper, "Probabilistic Traffic Forecasting with Dynamic Regression," has been accepted to Transportation Science!

[250225] New publication in Transportation Science

2025-02-25T00:00:00+00:00

I’m thrilled to share that our paper, “Scalable Dynamic Mixture Model with Full Covariance for Probabilistic Traffic Forecasting,” has been accepted for publication in Transportation Science!

This work was part of my postdoctoral research at McGill University in collaboration with Professor Nicolas Saunier, Mr. Vincent Zheng, Professor Martin Trépanier, and Professor Lijun Sun.

In this study, we introduce a Dynamic Mixture Model with a Full Spatiotemporal Covariance Matrix to better capture the time-varying distribution of spatiotemporal characteristics in network traffic data.

To overcome the limitations of conventional MSE/MAE-based training methods, which assume an independent and isotropic Gaussian distribution, we developed a Mixture Density Network-based training method with time-varying mixture weights. Additionally, we estimate the full spatial and temporal covariance matrix during model training—while ensuring scalability to handle large spatiotemporal datasets efficiently.

Preprint available at: https://lnkd.in/edDfXs5y

[250107] Transportation Research Board Annual Meeting 2025 presentation

2025-01-07T00:00:00+00:00

I will be presenting "Weaver: A Spatio-Temporal Deep Learning Model Architecture Based on the Mixed Kronecker Matrix-Vector Identity" at Transportation Research Board Annual Meeting (TRBAM 2025)

[240903] TRC-30 Conference presentations

2024-09-03T00:00:00+00:00

I will be presenting two papers at TRC-30: (1) Evaluating UAM Route Feasibility in Terminal Airspace via Probabilistic Aircraft Trajectory Prediction, and (2) Multi-level Traffic Simulation with Dynamic Simulation Level Assignment for Urban Network