Submitted by UF/IFAS Institute of Food and Agricultural Sciences — Artificial intelligence (AI) is creating new inroads into technology, conversation and publishing. But what about science and, specifically fighting mosquitos and killing weeds while preserving fruits and vegetables?
Researchers from the UF/IFAS Florida Medical Entomology Laboratory (FMEL) and the Florida Museum of Natural History are working to use AI to improve and revolutionize the precision of mosquito control. Their research aims to make it easier to apply larvicide effectively but in targeted areas, meaning less larvicide gets used where it isn’t needed, benefitting mosquito control programs through cost-savings and environmental protection.
Assistant Professor Lindsay Campbell of FMEL and Professor Robert Guralnick, the Curator of Biodiversity Informatics at the Florida Museum of Natural History, are researching the development of precision larvicide applications, using AI and advanced geospatial technologies such as LiDAR as a new tool to help target mosquito breeding areas with more accuracy than current methods, Campbell said.
“Precision larvicide application is about combining knowledge from mosquito control programs about highproducing larval habitats with technologies and AI to pinpoint areas where larvicides can be applied more effectively,” Campbell said. “By doing so, we can help reduce costs for mosquito control programs, slow the development of insecticide resistance and minimize environmental impact.”
The project uses AI to model mosquito larval distributions and predict optimal locations for larvicide applications.
“Recent advances in AI allow us to create models that are adept at filtering out noise and outliers, making them highly effective in predicting mosquito habitats,” Guralnick said. “These models can reveal complex relationships between preferred larval habitat and environmental factors that were previously difficult to detect.”
A key part of the research involves inputting hydrological data derived from LiDAR into the AI models to get detailed insights into how water collects on Florida’s coasts. The goal of using this information in the model is to predict with high accuracy where mosquito breeding sites are located.
At the same time, UF scientists hope to eventually help growers unleash surgical strikes on weeds – without vanquishing their fruit. Additionally, robots may replace tractors as the means for delivering the spray in the field. Getting rid of weeds is critical for growers in Florida’s $300 million annual strawberry industry because weeds hinder fruit yield.
Together, Boyd and Schumann—a professor of soil and water sciences at the Citrus Research and Education Center in Lake Alfred—are using artificial intelligence to detect and identify weeds within a crop canopy. The technology also can do the reverse: detect the canopy and spray everything else.
Boyd and his team trained AI programs to scan for weeds in the images collected by a camera. Once detected, a computer sends a signal to the spray system to spray the herbicide only on the weeds.
“This is done while moving through the field on a tractor, so you have to be able to detect weeds and hit a moving target,” Boyd said. No easy task, but it’s worth it. Boyd’s team is also working to improve spray precision at higher tractor speeds.
While the research remains in the preliminary stages, Boyd wants to reduce the amount of herbicide spray strawberry growers use. That saves farmers money and helps cut unwanted chemicals from getting into the environment.
Asked for cost-savings estimates, Boyd gave this example: If you shoot low and anticipate a grower using an inexpensive herbicide that costs $30 per acre and you drop the amount of spray by 50%, you save $15 per acre.
“That doesn’t sound like much, but if you have 100 acres, that is $1,500,” Boyd said. “Then add in the costs of the number of times you have to refill the herbicide tank that you’ll need in order to spray the whole field Now, UF/IFAS researchers are designing a machine that can surgically eliminate weeds while preserving the tomatoes and peppers around them utilizing artificial intelligence (AI) and sensor fusion. Vinay Vijayakumar, a doctoral student, created the smart-spraying system that targets weeds on raised beds of soil and in row middles, the areas between rows of plastic beds. The smart machine’s movement is currently operated by remote control, but future advancements will enable it to function as a fully autonomous system, like a robot. The smart-spraying system can also be mounted on the back of a tractor or any other farm vehicle.
So far, the prototype properly distinguishes 98% of the pepper and tomato plants and 85% of the weeds, said Vijayakumar. The machine-vision system then takes information from the camera and relays it to the spraying system, which sprays the detected weeds, while avoiding the vegetable crops.
“With this setup, we aim to achieve high spraying accuracy to cover the entire raised bed width while maintaining a spray strong enough to prevent drift and reduce off-target spray,” Vijayakumar said.
