PhD Student · University of South Florida · MOTIF-Lab

Yuhang Wang 王雨航

Vision–language models that decide when to alert the driver.

View Projects Google Scholar 🤗 Hugging Face Email
  • IEEE/CVF WACV 2027
  • CoRL 2026
  • IEEE ITSC 2026
  • IEEE ITSC 2025
  • TRB 2026
  • SMI 2024 · Best Paper
  • CoRL 2026 · Travel Award

01At a glance

My PhD builds vision–language models, large-scale real-world datasets and benchmarks for safer driving automation.

  • 7Projects6 datasets & benchmarks, 1 on-road system
  • 6Selected papersWACV 2027 · CoRL 2026 · ITSC 2026 · ITSC 2025 · TRB 2026 · SMI 2024
  • 975 hNaturalistic driving in DriveDNA4,121 drives · 465 drivers
  • 192,892Labeled ticks in VLAlert-Bench13,534 videos · 6 datasets
  • 2Conference awardsBest Paper SMI 2024 · Travel Award CoRL 2026
  • Multimodal LearningVideo, CAN, radar & language — fused for driving understanding
  • Human-Centered AIAlerting, trust and shared control that drivers accept
  • Driver Behavior ModelingTakeovers, driving style and body motion at scale
  • Autonomous DrivingSafety of production ADAS & automated vehicles

02Projects

Featured Projects

Six datasets and benchmarks and one on-road system, all built from real drivers, real vehicles and real roads.

All projects

IEEE/CVF WACV 2027

DriveMotion

Driver body-motion forecasting

400 h of in-cabin motion as 133-keypoint skeletons at 10 Hz from 360 drivers — released skeleton-only.

Open project

CoRL 2026 Travel Award

VLAlert

Observe before you alert

Decides when to stay SILENT, OBSERVE or ALERT; VLAlert-Bench: 192,892 labeled ticks from six datasets.

Open project

arXiv 2026 · under review

TriDrive

Joint driver–vehicle–road modeling

Forecasts driver kinematics, vehicle dynamics and road demands in real time on comma four; validated with 14 drivers on the road.

Open project

IEEE ITSC 2026

ADAS-TO

Human–ADAS takeover transitions

15,659 real-world ADAS takeover clips from 327 drivers across 22 brands, with vision-language annotations.

Open project

IEEE ITSC 2025

OpenLKA

Lane keeping on real roads

Lane Keeping Assist from 62 production vehicle models — decoded CAN logs paired with synchronized video.

Open project

arXiv 2026

BATON

Bidirectional control transitions

781 routes, 173 drivers, 204.9 h of driving-automation use; takeover scored as detection and as anticipation.

Open project

arXiv 2026

DriveDNA

Driving-style identification

4,121 drives from 465 drivers, 975 h — isolating driver-specific behavior from vehicle, route and environment.

Open project

Index

All projects

Every dataset, benchmark and system — with papers, code and data access in one place.

  • DriveMotion WACV 2027
  • VLAlert CoRL 2026
  • TriDrive arXiv 2026
  • ADAS-TO ITSC 2026
  • OpenLKA ITSC 2025
  • BATON arXiv 2026
  • DriveDNA arXiv 2026
Open the index

03Publications

Selected Publications

Five recent papers, newest first. Preprints and the complete list are on the publications page.

All publications
  1. WACV 2027

    DriveMotion: A Large-Scale Multi-Source Benchmark for Driver Motion Sequence Modeling and Forecasting

    Yuhang Wang, Chuheng Wei, Jingxin Yang, Xishun Liao, Hao Zhou

    IEEE/CVF Winter Conference on Applications of Computer Vision (WACV 2027) · Lake Buena Vista, Florida, United States. January, 2027.

    ProjectarXivData

  2. CoRL 2026 · Travel Award

    Observe Before You Alert: Adaptive Driver Alerting with Vision–Language Models

    Yuhang Wang, Lingyao Li, Hao Zhou

    Conference on Robot Learning (CoRL 2026). CoRL 2026 Travel Award.

    ProjectPDFarXivData

  3. ITSC 2026

    ADAS-TO: A Large-Scale Multimodal Naturalistic Dataset and Empirical Characterization of Human Takeovers during ADAS Engagement

    Yuhang Wang, Yiyao Xu, Jingran Sun, Hao Zhou

    IEEE Intelligent Transportation Systems Conference (ITSC 2026)

    ProjectarXivDataCode

  4. ITSC 2025

    OpenLKA: An Open Dataset of Lane Keeping Assist from Production Vehicles Under Real-World Driving Conditions

    Yuhang Wang, Abdulaziz Alhuraish, Shengming Yuan, Hao Zhou

    IEEE Intelligent Transportation Systems Conference (ITSC 2025) · Gold Coast, Australia. November, 2025. Journal extension under review at Accident Analysis & Prevention.

    ProjectIEEE XploreCodeData access

  5. SMI 2024 · Best Paper

    Computational design of custom-fit PAP masks

    Yukun Lu, Yuhang Wang, Peng Song, Hang Siang Wong, Yingjuan Mok and Ligang Liu

    Shape Modeling International (SMI 2024) · Wayne State University, United States. July, 2024.

    DOISMI 2024

04News

News & Updates

Acceptances, releases and awards — newest first.

  1. 🏆 Awarded a CoRL 2026 Travel Award ($1,000) for Observe Before You Alert at the Conference on Robot Learning. [Awards]

  2. 🎉 DriveMotion: A Large-Scale Multi-Source Benchmark for Driver Motion Sequence Modeling and Forecasting is accepted by WACV 2027! See you at Disney Springs, Florida, in January. [arXiv] [Data]

  3. 🎉 Seven papers accepted for presentation at the TRB 2027 Annual Meeting — five first-author (CockpitWM, Observe Before You Alert, BATON, Human Takeovers in Level-2 ADAS, DriveDNA) and two co-authored. See you in Washington, D.C.!

  4. 🚗 New project page: TriDrive — Joint Driver, Vehicle, and Road Modeling for Human-Centered Driving, a world-model-based driver-monitoring system running in real time on comma four, validated in a 14-participant on-road study. [Explore]

  5. 🎉 Observe Before You Alert: Adaptive Driver Alerting with Vision–Language Models is accepted by CoRL 2026! [PDF]

  6. 📦 DriveMotion — a multi-source, skeleton-only benchmark for driver motion forecasting (400 h, 360 drivers) — is on arXiv and Hugging Face.

15 earlier updates

  1. 🎉 Three papers accepted by ITSC 2026: ADAS-TO, Cut-In Gap Acceptance Toward Autonomous vs. Human-Driven Vehicles, and A Closed-loop, State-centric, Multi-agent Framework for Passenger Load Estimation.

  2. 🧬 DriveDNA — driving-style identification without the shortcuts (4,121 drives, 465 drivers, 975 h) — is on arXiv.

  3. 🔁 BATON — a multimodal benchmark for bidirectional human–automation control transitions — is on arXiv (v2, Jul 2026: 781 routes, 173 drivers, 204.9 h).

  4. 3 papers accepted by TRB Annual Meeting — see you in D.C.!

  5. From OpenLKA to LKAlert abstract accepted by INFORMS 2025 (Intelligent Transportation Systems and Cybersecurity session), Atlanta.

  6. OpenLKA is accepted by ITSC 2025.

  7. Extended abstract accepted by the Second USF Artificial Intelligence + X Symposium (poster).

  8. Computational design of custom-fit PAP masks won the Best Paper award at SMI 2024.

  9. Received a PhD offer from the Civil Engineering department at the University of South Florida!

  10. Computational design of custom-fit PAP masks accepted for presentation at SMI 2024.

  11. Started as a visiting student at SUTD CGL Lab, advised by Prof. Peng Song.

  12. 🥈 Silver Medal at CCPC Guilin — Team 廉颇老矣，尚能夺金? · rank 47.

  13. Joined ZelosTech's PnC group as an algorithm engineering intern.

  14. 🥇 Gold Medal at the ICPC National Contest — Team Tops dogs like real party · rank 16.

  15. Joined the team supervised by Prof. Xiaolong Xu.

05About

About Me

Yuhang Wang
05 · Tampa, FL

I am Yuhang Wang (Henry, 王雨航), a PhD student in Civil Engineering Transportation at the University of South Florida, working as a research assistant at the MOTIF-Lab advised by Prof. Hao Zhou. I previously spent a year with Prof. Peng Song at the Singapore University of Technology and Design.

My research applies deep learning and the capabilities of self-driving vehicles to transportation problems — building the large-scale, real-world datasets the field is missing, and pursuing human-centered driving automation that earns drivers' trust. If any of this interests you, I would love to chat and collaborate: yuhangw@usf.edu.

Vision–Language ModelsDatasets & Benchmarks Autonomous Driving SafetyMultimodal Learning Human-Centered AIDriver Behavior Modeling
  1. Aug 2024 – Present

    University of South Florida — Transportation PhD student, advised by Prof. Hao Zhou

  2. Oct 2023 – Jul 2024

    Singapore University of Technology and Design — Visiting student, advised by Prof. Peng Song

  3. Sep 2020 – Jun 2024

    Nanjing University of Information Science and Technology — BEng in CS, advised by Prof. Xiaolong Xu