Researchers train robot dog to fight invasive fire ants

A multidisciplinary research team based in China and Brazil has used a dog-like robot and AI to create a new way to find fire ant nests. Published in the journal Science of Pest ControlThe study highlights how a “CyberDog” robot integrated with an AI model can automate the identification and control of red imported fire ants (RIFA), a globally destructive pest.

Field tests conducted by the researchers reveal that the robotic system can significantly outperform human inspectors, identifying three times more RIFA nests with greater accuracy. Eduardo Fox, a postdoctoral researcher at the State University of Goiás in Brazil and corresponding author of the study, explained the motivation behind this pest management approach.

“Fire ant nests are difficult for untrained personnel to identify and confirm in the field, and searching large areas can be time-consuming and exhausting in the hot sun. A robot could automatically locate nests without requiring specially trained people and operate at different times of day, regardless of temperature conditions,” Fox said.

Controlling an invasive species

RIFA is one of the most destructive pests in the world. Accidentally introduced into the United States in the 1930s, it has since spread as an invasive organism to many regions, including China, Japan, and Europe, causing significant environmental damage and economic losses.

Speaking about the importance of detecting RIFA nests, Fox noted, “RIFA proliferate rapidly in established areas, displacing local wildlife by outcompeting and eliminating susceptible species. Small vertebrates, such as birds and reptiles, are particularly vulnerable. Additionally, RIFA are associated with important agricultural pests such as scale insects and can damage certain plants, exposing them to pathogens.”

Conventional approaches to control RIFA populations involve the use of pesticides, which can harm local ecosystems. To implement effective, targeted RIFA control strategies that minimize harm to native species, extensive population monitoring is essential.

Train the dog

The team used Xiaomi’s CyberDog robot, integrating it with a machine learning model trained on a comprehensive dataset of over 1,100 RIFA nest images. This approach achieved a nest detection accuracy rate of over 90%.

The researchers conducted rigorous field tests to measure the effectiveness of the system. The CyberDog was programmed to press down on the nest with its front paw: when a mound of fire ants is disturbed, the workers rush out of cracks and openings, displaying aggressive behavior. This, the researchers say, is essential for diagnosing active mounds from abandoned nests and for avoiding false positives with mounds inhabited by other species.

Hualong Qiu, a researcher at the Guangdong Academy of Forestry in China and corresponding author of the study, explained: “A group of students received official standard training for quarantine inspectors and were tasked with locating fire ant nests in an open field. Subsequently, the AI-trained robot was challenged on the same field and the performances of the students and the robot were compared.”

Despite the promising results, researchers acknowledge several challenges in scaling the technology.

“The main limitations to using robots are their battery life, which lasts about 30 minutes, and the high cost of acquiring more agile and efficient models,” noted Zheng Yan, a researcher at Lanzhou University in China and corresponding author of the study.

“Currently, it is still more expensive to use the robotic system than the traditional approach, but we believe that production costs can be optimized over time,” he said.

The results of the study could have a significant impact on pest control policies and public awareness.

“In addition to being versatile machines for navigating urban environments, robot dogs have attracted a lot of public attention. Fire ants pose a serious threat in China, but most people are unaware of the dangers posed by invasive fire ant nests in public spaces. Therefore, sightings of robots tracking fire ant nests are likely to captivate the public and raise awareness of the presence of fire ants,” Yan said.

Provided by the Chemical Industry Society