Welcome
At the Yang Research Group at Johns Hopkins University, we build AI copilots that help people make complex decisions. We combine scientific theory, domain knowledge, and physical constraints to help these systems reason through engineering problems, run simulations, and use tools. We study how copilots can learn from people’s choices and feedback, adapt to their goals and values, and help them judge when to trust a recommendation. We also design adaptive mixture-of-experts systems that match computing effort to the task to reduce response times, energy use, and deployment costs. Transportation and public health are our main application areas, where we test how these copilots can strengthen human judgment in practice.
News
GenAI copilots for transportation decision-making
With support from the NSF and National Academies, Dr. Hao Frank Yang is developing engineered AI copilots that combine transportation evidence, established models, and simulation tools to help agencies explore alternatives and make informed planning decisions.
Read more →
SafeRoads AI advances road safety with Google.org support
As technical lead of SafeRoads AI, Dr. Hao Frank Yang is helping develop a platform that combines multimodal municipal data and agentic AI to support road safety decisions. Supported by Google.org, the project works with city partners to identify risks and evaluate safety improvements with human oversight.
Read more →
NSF award supports AI-driven traffic safety research
Dr. Hao Frank Yang has received an NSF CDS&E award to develop the SafeTraffic Decision Copilot. The project combines multimodal foundation models, data utility theory, and uncertainty-aware analysis to predict crash risks, explain contributing factors, and recommend targeted safety interventions.
Read more →
Paper published in Transportation Research Part C
Hopkins engineers put large language model agents behind the wheel to study how AI agents reason in driving scenarios.
Read more →
NCHRP project awarded on generative AI for transportation planning
Dr. Hao Frank Yang will lead a Johns Hopkins research project developing a primer for using generative AI models in transportation planning.
Read more →
Three papers accepted to NeurIPS 2025, including two Spotlights!
Three of our papers have been accepted to NeurIPS 2025, including two selected for Spotlight presentations. Congratulations to all the authors!
Read more →
Paper published in IEEE TPAMI
A research team led by Dr. Hao Frank Yang developed an EMOS framework to improve autonomous driving safety and efficiency.
Read more →
Paper published in Nature Communications
Our paper on SafeTraffic Copilot has been published in Nature Communications and featured by the JHU Hub. Congratulations to the authors!
Read more →