Grape breeders across the country are working to develop varieties that won’t fall victim to powdery mildew, a fungus that threatens grapes just about anywhere they are grown, and now artificial intelligence is speeding up the process, a USDA scientist says.

Until recently, the reseasrch group that Lance Cadle-Davidson is part of relied on humans to sit at microscopes for hours looking at fungal spores and how far they grew. They realized that some of their processes could be automated.

Enter BlackBird, a robot developed in collaboration with engineers and computer scientists at Cornell University that can capture thousands of images, feed them into a neural network and identify the information they’re looking for even more accurately than the human technicians could.

Cadle-Davidson, a research plant pathologist with USDA's Agricultural Research Service in Geneva, N.Y., said “it does take us a lot of time” to teach the robot what it needs to know, time that's spent in front of a large computer screen, not bent over a small microscope.

Interested in more coverage and insights? Receive a free month of Agri-Pulse West

Once the analysis identifies the samples most likely to be resistant to powdery mildew, Cadle-Davidson’s lab provides plant breeders with DNA tests they can use on-site to pick the best candidates for their breeding programs. Ultimately, the goal is that public or private grape breeders will be able to launch commercial varieties that are resistant to the fungus. That could reduce dependence on pesticides, which Cadle-Davidson said are currently sprayed as often as 10 to 15 times per year to protect vines.

 ARS is spending $100,000 to bring BlackBird to other field offices so it can help with any crop susceptible to powdery mildew, which includes strawberries, hops and many others, he said. 

“We’ve developed a framework,” he said. “We want to generalize it even further.”

For more news go to