NSF grant advances safer self-driving cars
A University of Arizona connected vehicle testbed collects sensor data vital to developing and testing cooperative perception technology for autonomous vehicles.
Two engineers in the School of Electrical, Computing, and Software Engineering have received $625,000 from the National Science Foundation to develop faster, more reliable communication systems for self-driving cars – technology that could help prevent accidents and save lives.
The award is part of an $880,000 grant shared with North Carolina State University. Ming "Fred" Li, professor and University Distinguished Scholar in ECSE, is the project's principal investigator, along with Regents Professor Marwan Krunz, Edward and Maria Keonjian Endowed Chair as the co-PI. The two are collaborating with Wujie Wen, associate professor of computer science at NCSU.
Solving a hidden safety risk
Connected and autonomous vehicles rely on a cooperative perception process that shares real-time data with nearby cars and roadside sensors to detect hazards. The technology allows cars to effectively see around corners, through fog and past large trucks blocking the road ahead.
However, that safety benefit comes with a catch. The cameras, lidar and radar sensors on connected vehicles generate massive amounts of data.
"Raw sensor data is too large and can easily overwhelm network bandwidth, causing delays that compromise safety," Li said. "Our goal is to send only the essential, safety-critical information vehicles need to make split-second decisions."
Smart communication
Li and Krunz are developing a "semantic" communications framework. Rather than transmitting entire raw data streams, the system uses machine learning to identify information critical to safe driving decisions.
"The goal isn't just to move data faster. It's to make sure the right information reaches a vehicle at the right moment, every time," Li said.
Their model will compress and combine sensor data, quickly prioritizing safety-critical data to meet the strict speed and reliability demands of autonomous driving.
From lab to road
The NSF grant will allow the team to test their framework on connected and autonomous vehicle testbeds.
“That real-world testing is essential to making these systems dependable enough for commercial deployment," said Krunz.
The team is already partnering with industry leaders Denso, an automotive tech company, and Qualcomm, a wireless communication company.
"Collaborating with industry from day one means our results won't just live in a lab," Krunz said. "They'll be shaped by the standards and constraints of real deployment."
The grant also funds hands-on research experience across grade levels. Graduate students will lead core technical work, including algorithm design and system integration, while undergraduate researchers will conduct testbed experiments and gather sensor data. K-12 students will have the opportunity to be involved through university STEM outreach events and demonstrations.
"This project reflects exactly the kind of research our school was built to do – solving real safety problems while training the next generation of engineers," said Michael Wu, ECSE head and Thomas R. Brown Leadership Chair.