Robots + Drones + AI = Hardier Tomatoes?

Date:12-04-2024

Each year, plant diseases take a bite out of agricultural yields and profits, causing estimated global agricultural losses of $30 billion to $50 billion. At North Carolina State University, experts in optical engineering, data analytics, and robotics are working with plant breeders and pathologists to get around a bottleneck that hinders the development of disease-resistant crops.

The team recently won a three-year, $740,000 grant from the U.S. Department of Agriculture’s National Institute of Food and Agriculture to develop a software and hardware package to help tomato breeders score diseases in the field and accurately measure their severity.

That information will be key to breeders’ developing higher-yielding, disease-resistant varieties.

The idea, says principal investigator Lirong Xiang, ″is to combine drone-based field scouting with ground-vehicle scouting.″

Bringing diverse expertise to bear

Xiang, a robotics expert, is an assistant professor in NC State’s Department of Biological and Agricultural Engineering. Other team members include tomato breeder Dilip Panthee, NC State Extension plant pathologist Inga Meadows, electrical and computer engineering professors Cranos Williams and Michael Kudenov, and remote sensing expert Leila Hashemi-Beni of North Carolina A&T State University.

Xiang, Williams, and Kudenov are affiliated with the N.C. Plant Sciences Initiative. Panthee, of the Department of Horticultural Science, is stationed at the Mountain Horticultural Crops Research and Extension Center in Mills River, and Meadows, of the Department of Entomology and Plant Pathology, works at the Mountain Research Station in Waynesville. Hashemi-Beni is with N.C. A&T’s College of Science and Technology.

Advancing Panthee’s efforts

In his plant breeding program, Panthee combines conventional and molecular approaches to developing cultivars and breeding lines that resist diseases.

His success depends, in part, on being able to accurately detect and quantify bacterial and fungal diseases in his test plots, he says. Right now, Panthee trains workers to scout for disease multiple times throughout the summer, scoring symptoms based on their severity.

What they find helps Panthee select plants that might make good parents for disease-resistant offspring – ones he’ll continue to use in his breeding program to create cultivars that are not only disease-resistant but have other traits like flavor and texture that consumers prefer.

Tomato plant with foliar disease highlighted by artificial intelligence models.

The problem, Panthee says, is that the system is labor-intensive and subjective. To get around these challenges, the researchers plan to train computers to do that work faster and more precisely. Robots and drones will take photographs, and computers will generate data-rich maps that pinpoint plants with disease and indicate the severity of the disease.

Using drones is fast, Xiang says, but the researchers think that adding robots that move along the ground could lead to more accurate assessments.

″Many diseases develop from the bottom of the canopy to the top, and there’s a possibility that the drone-based imaging cannot capture the disease symptoms that are under the canopy and even inside the canopy,″ Xiang says.

Toward sustainable solutions for higher yields

Right now, Xiang and her students are developing a robotic system for collecting images, with the goal of deploying the system at the Mills River and Waynesville stations next summer.

″By the time the project ends in 2027, we hope to have a software and hardware package ready for breeders to use,″ she says.

Meadows, the team’s disease expert, sees the development of varieties with desirable fruit qualities and disease resistance as the most environmentally and economically sustainable way of enabling farmers to meet the growing demand for tomatoes.

She explains, ″The robot will allow us to observe differences in disease resistance among breeding lines that the human eye cannot see and hopefully identify some durable disease-resistant varieties for growers.″