I’ve covered research papers that involved the red imported fire ant (RIFA), Solenopsis invicta, in previous articles and podcast episodes, so I’ll keep my background information relatively brief¹. The important cliff notes are that RIFA is an incredibly damaging invasive pest that wreaks havoc on non-native ecosystems. It’s native to South America and has slowly made its way around the world thanks to various methods of human mediated transport. Workers are incredibly aggressive when foraging for food and defending territory, allowing this species to easily outcompete other ants and organisms for resources. The RIFA also packs a painful sting that can be fatal to anyone with an allergy to insect stings. To convert their destruction into cold hard cash, the RIFA ranks as the 5^th costliest invasive species globally, accounting for approximately $17 billion in economic damages and management costs over the past few decades (1970 to 2017)².  

Stopping the spread of the RIFA is critical to minimizing ecological and economic damages as well as protecting public health. But that’s easier said than done. Despite our best efforts to contain it, the RIFA has spread across most of the southern US since its initial introduction in the 1930’s. Early detection is critical to stopping or slowing the RIFA’s expansion, but finding stow-away ants in transported goods can be tough. In the past, detection methods have included physical inspections of transported goods and visual surveillance to locate mounds and foraging workers in the field. Both options can be effective but there is always the potential for human error as reproductives or small nest fragments hidden in transported goods can easily be missed, and smaller mounds may be overlooked.  

Researchers at China’s Lanzhou University set out to improve fire ant surveillance by removing the human element all together with the help of advanced robotics and artificial intelligence (AI). The team trained an open-source AI system to identify RIFA nests by uploading over 1,000 images taken of nest mounds from various angels and conditions. Once the AI program was ready, it was installed onto a state-of-the-art robotic Cyberdog equipped with sophisticated sensors, cameras, thermal imaging devices, chemical detectors, and other technology that allow the robot visually process the surrounding environment and compare that data to the upload of fire ant mound images.  

To test the performance of the ant-detecting robot k9 against humans, researchers gave the Cyberdog and three human surveyors 10 minutes to inspect a 300-square-meter nursery for RIFA mounds. At the end of the survey period, the Cyberdog found three times more nests than its human counterparts. In addition, the cyberdog missed fewer nests and had a lower false detection rate than the human inspectors. 

 

 

Although this technology is still in the early development phase and likely comes with a heavy price tag, this study gives great promise to what the future of pest surveillance and invasive species  management might look like. Imagine a not-so-distant future where you show up to your client’s home to perform their regular service. Before going inside to chat with your client, you deploy your CyberDog Pest Detector 1000 (I totally made this name up by the way) to inspect the exterior of the structure and surrounding property. By the time you’ve finished resolving any interior issues, the CyberDog has prepared a detailed report of pest activity and has mapped locations and treatment recommendations for each incidence logged. This is all hypothetical of course, but based on the direction that technology and AI are heading, this may not be as far “fetched” as it sounds.   

Article by Mike Bentley, PhD, BCE

References

Xin Su, Guijie Shi, Jiamei Zhong, Yuling Li, Wennan Dai, Guohua Xu, Eduardo G. P. Fox, Hualong Qiu, Zheng Yan. Use of Artificial Intelligence for Automated Detection and Surveillance of Red Imported Fire Ants Nests. bioRxiv 2023.05.26.542461; doi: https://doi.org/10.1101/2023.05.26.542461

Podcast