As automated vehicles become more sophisticated, effective interaction depends on more than following simple commands. A vehicle may need to interpret a person’s broader intent while simultaneously accounting for traffic conditions, roadway rules, vehicle capabilities and other safety constraints. Transportation engineers at the University of Utah are bringing those elements together to develop AI-enabled control methods that respond to human goals without compromising the safeguards required in real-world transportation systems.

Turning that goal into reality requires substantial computational power. Modern AI systems must be trained and evaluated using large datasets and thousands of simulated scenarios before researchers can understand how they will respond under varied—and sometimes unexpected—conditions. Access to that level of computing is limited, making it an important research resource in its own right.

A new national award will give University of Utah researchers the computing capacity to take on that challenge.

Chenxi “Dylan” Liu and Cathy Liu have received 9,200 hours of advanced GPU computing through the National Science Foundation-led National Artificial Intelligence Research Resource Pilot. The award will support their work at the intersection of transportation engineering, artificial intelligence and human–vehicle interaction.

Their project, “INTERACT: Intent-Driven Human–Vehicle Interaction with Constraint-Aware Control,” explores how intelligent vehicles can better understand what people want them to do—and translate those intentions into actions that remain safe and technically feasible.

Through the NAIRR Pilot, Liu and Liu will have access to 9,200 hours on NVIDIA A100 graphics processing units, or GPUs. These high-performance processors can handle many computations simultaneously, making them especially valuable for training AI models and testing complex control systems.

Unlike a traditional research grant, the award provides computing capacity rather than direct funding. For computationally intensive research, however, access to that capacity can be just as consequential. The allocation will allow the team to conduct experiments at a scale that would be difficult to achieve using conventional computing resources.

Led by the U.S. National Science Foundation in partnership with other federal agencies and nongovernmental organizations, the NAIRR Pilot is building a shared national infrastructure for responsible AI research and innovation. The initiative connects researchers and educators with advanced computing, datasets, software, models, training and technical support that might otherwise be beyond their reach.

Its goals include expanding participation in AI research and supporting applications in areas such as health, environmental sustainability and infrastructure. According to NSF, these resources are intended to broaden access to the tools needed both to advance AI research and to conduct research that employs AI.

For the Department of Civil & Environmental Engineering, the award reflects a growing research presence where civil infrastructure, transportation and artificial intelligence converge. It also demonstrates that AI innovation is not limited to traditional computing disciplines. Civil engineers are increasingly helping determine how intelligent systems interact with people, vehicles and the built environment—and how those technologies can be deployed responsibly.

By earning access to nationally competitive AI resources, the INTERACT team will be able to accelerate its work toward vehicle systems that are not only more capable, but also more intuitive, responsive and safe.