Researchers developing AI-powered drones to protect peanut farms

Feral hogs in Georgia cause an estimated $150 million in agricultural damage each year, according to the U.S. Department of Agriculture. To address this growing threat, Kennesaw State University researchers are launching an AI-powered drone system to autonomously protect peanut fields from nighttime wildlife intrusion and reduce costly crop losses.

Supported by a $25,000 Georgia Peanut Commission grant, the one-year Night Owl project is a collaboration between Kennesaw State assistant professors Taeyeong Choi and Yan Fang, and University of Georgia agricultural economist Ford Ramsey. The team's mission is to create a low-cost, AI-driven solution that deters destructive wildlife from farmland without the need for traditional fencing or human patrols.

"The idea came from conversations with local farmers, who shared that wild animals were a bigger problem than pests, especially at night," said Choi, who teaches information technology in KSU's College of Computing and Software Engineering (CCSE). "Traditional deterrents like fences are too expensive to install and maintain, so they often rely on staying up late with dogs to chase the animals off. That's simply not sustainable."

Night Owl uses stationary cameras to detect animal movement in fields and dispatches autonomous drones equipped with infrared sensors. These drones fly to the location, flash lights or emit noise to scare off the intruders, and then return to a centralized charging post, all without human intervention.

"It's not only more affordable than fencing, but also smarter," said Fang, who teaches electrical and computer engineering in KSU's Southern Polytechnic College of Engineering and Engineering Technology. "Farmers will know if the system is working in real time. More importantly, it operates within Georgia laws. We are not harming the animals, just deterring them."

Training the AI system to perform well in low-light conditions is one of the project's biggest challenges. The team will collect and manually annotate images from real-world environments, likely beginning at the KSU Field Station, to help the software accurately identify animals in the dark.

From a hardware perspective, the team must ensure that the drones can function reliably under variable weather conditions and complete their missions efficiently before returning to charge. Fang, whose research includes energy-efficient AI and robotics, noted that optimizing energy use will be a challenge during the prototype phase.

The project is being created within CCSE, which, along with the university, supports interdisciplinary initiatives that apply emerging technologies to real-world challenges. Once the project exits the testing phase, the team envisions scaling the technology beyond peanut farms to other crops and regions.

Ramsey's economic analysis will be critical in demonstrating how Night Owl translates to real-world value. By comparing farms that use the system with those that do not, he will help quantify cost savings and crop protection in terms that resonate with farmers and funders alike.

"We're not replacing farmers," Choi said. "We're giving them a tool that lets them rest at night while knowing their fields are protected. The feedback we've received from farmers so far has been overwhelmingly positive."

Provided by Kennesaw State University