The Department of Defense has named Dr. Anthony Fragoso, a research physicist at the U.S. Army Engineer Research and Development Center's (ERDC) Cold Regions Research and Engineering Laboratory (CRREL), a 2025 Laboratory-University Collaboration Initiative (LUCI) fellow.
LUCI provides funding of up to $200,000 per year for three years for research collaborations between fellows and prominent DoD-funded academic researchers – either recipients of the Vannevar Bush Faculty Fellowship or principal/co-principal investigators under the Multi-Disciplinary University Research Initiative.
The LUCI fellowship will pair Fragoso, whose research focuses on developing AI and autonomy for extreme environments, with Prof. Richard Murray, a control theorist and the Thomas E. and Doris Everhart Professor of Control and Dynamical Systems and Bioengineering at the California Institute of Technology.
Their project – Automated Test Generation for Arctic AI and Autonomy – will research the fundamentals of knowing when AI works and how it breaks, exploring the algorithms, mathematics and theory behind it.
While Fragoso says a lot of research has been conducted on how AI performs in structured, relatively tame environments – think whether or not an autonomous car will know how to behave at a four-way stop sign or stop for a school bus with flashing lights – much less is known about how AI performs in unstructured, extreme conditions like those encountered in the Arctic.
“I've always been fascinated with how adding winter and extreme cold changes everything,” says Fragoso. “CRREL exists because things break in the cold in unexpected, weird ways that have challenged armies and societies for millennia. Even though AI algorithms have accomplished so much so fast and have been a major technological breakthrough, right now they’re also unfortunately at the cutting edge of breaking in unexpected, weird ways that we cannot currently comprehend. This is an opportunity to start looking at why.”
Hard to catch on camera
Fragoso and Murray are interested in using formal methods to test AI in Arctic settings because, as Fragoso puts it, the Arctic is among the most taxing places to deploy AI.
“It is a totally unstructured place,” said Fragoso. “You can drive an autonomous over-snow vehicle pretty much anywhere, but also get into trouble pretty much anywhere. In the winter, everything is snow-covered and there's very little texture to work with, very different than a typical suburban streetscape.
“At the same time, even though there's a lack of structure and texture, the things that get you into trouble are difficult to notice and don't really arise anywhere else. You have to go there and see a few of them first to deal with them correctly. Things like sinking into a snow drift, getting stuck in melted permafrost, or hitting a buried obstacle, which are very hard to see in advance. It all depends on what's underneath it. This is tough when all you have is ‘snow is white’ and ‘grass is green.’”
And that, says Fragoso, is in the winter when things are relatively simple. It gets even more complicated as the Arctic warms seasonally.
"There is midwinter thaw, rain-on-snow and muddy conditions in the summer,” he added. “The Arctic has extreme conditions that break things but are hard to catch on camera.”
An investment that empowers our scientists
According to the DoD, the LUCI program fosters collaboration between DoD laboratory scientists and top university scientists funded by the DoD in priority defense research areas including applied mathematics, cognitive neuroscience, engineering biology, novel materials, quantum information science, and manufacturing science.
"The LUCI fellowship is an investment that empowers our scientists to push the boundaries of discovery while building strong ties with the academic community," said Dr. Jean-Luc Cambier, director of technical research programs in the Basic Research Office. "It provides our researchers with a unique opportunity to engage in exploratory research, pursuing transformative ideas that strengthen the department's scientific foundation and drive future innovation."
The selection process for the fellowship involved program officers from the U.S. Army, Navy, and Air Force, senior leadership from the Basic Research Office within the Office of the Under Secretary of Defense for Research and Engineering, and technical experts from partnering defense contractors. Out of 59 initial white papers, evaluators short-listed 22 proposals demonstrating strong potential for innovative, high-impact research.
Following a rigorous interview phase, the Basic Research Office selected 14 projects, represented by 20 principal and co-principal investigators from the military service laboratories.
“Tony is an excellent researcher whose work has major implications for DoD activities in the Arctic and other regions that are critical to the Nation’s defense,” said Dr. Melisa Nallar, chief of the Force Projection and Sustainment Branch at CRREL. "His dedication and innovative work are well-deserving of this prestigious fellowship, and I am thrilled to see his contributions recognized."