Every year, about 30 percent of the crops on Chris James’s Fresh Earth Farm in Hastings, Minnesota are lost to deer, who trek through and help themselves to whatever they find growing. To date, the best solution he had was to put up a fence blocking access to crops. But, James says, fences reduce usable farmland by about 20 percent, meaning farmers still end up losing a large portion of their yield.

This is one predicament the University of Minnesota is addressing by taking a lead role in the National AI Research Institute for Land, Economy, Agriculture & Forestry (AI-LEAF), funded by a $20 million federal grant. The institute is part of the National AI Research Institutes (NAIRIs), an initiative funded primarily by the National Science Foundation (NSF), which aims to promote nationwide collaboration in AI research. The mission of AI-LEAF is to create a new scientific discipline and innovation ecosystem, from studying how AI can be used to enhance the measurement of carbon and water flux to creating specialized field-to-market decision support tools.

“On our campus, the collaboration between agriculture and computer science, or AI, goes back many decades,” says Shashi Shekhar, a professor in the Department of Computer Science & Engineering and AI-LEAF’s principal investigator and director.

Securing the project’s five-year grant, which was awarded in 2023, was an extensive process. Researchers from the College of Science and Engineering and the College of Food, Agricultural and Natural Resource Sciences came together to craft a research idea that would spur advancements in multiple areas. The group assembled a team of partner universities for the project, including Cornell, Colorado State, Delaware State, North Carolina State, and Purdue, and surveyed farmers to ensure the project targeted areas that would be useful in practice.

The grant included three rounds of proposals: “The total documentation was almost 1,000 pages,” Shekhar recalls.

AI has transformative potential for farming. Farmers must make many decisions about land-management practices: whether to till, how much fertilizer to use, in which areas to plant seeds. Because there are a substantial number of economic and environmental factors to weigh, these conclusions can take time to calculate and still might only yield an educated guess. The consequences can be severe. “[Farmers] may lose a whole season. They may have to take on a lot of debt or go out of business,” Shekhar says.

Project leaders are incorporating input from farmers. James suggested placing cameras around the field, which can detect motion and direct drones toward intruders. These drones, programmed to recognize the difference between deer and humans, are designed to drive away deer using a variety of methods. “It could be lights, could be noises, could be just flying towards the deer,” says James. What matters is variation—deer tend to fear only what they don’t understand, meaning they might learn to ignore the drones if they became too predictable.

Farmers like James are helping project leaders design solutions for all dimensions of fields. “I think [the team] might’ve had the impression that farms are just big open spaces,” says James. In reality, his property is dotted with obstacles—trees, power lines, even a greenhouse—that a drone would need to know how to avoid. Now, a team of University of Minnesota students, led by AI- LEAF scholar Maria Gini, is testing a drone designed to do this.

“Minnesota agriculture is a significant part of our economy,” adds Shekhar. “We are very honored to have this opportunity.”

If you liked these stories, Minnesota Alumni magazine publishes four times a year highlighting U of M alumni and University activities. Early access to stories and a print subscription are benefits of being an Alumni Association member. Join here to receive a printed copy at home.