IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Washington Universities to Lead AI Teams in Ag, Engineering

Washington State University and the University of Washington will head new research geared toward using artificial intelligence to solve problems associated with climate change and real-time machine learning.

Washington State University researchers are briefed on drones outfitted with technology to collect multispectral and thermal imagery data at a "smart farm" in Yakima County, Wash.
(Robert Hubner/WSU)
The University of Washington and Washington State University will lead two new research institutes in developing artificial intelligence to solve complex problems in engineering and agriculture, using $40 million from the National Science Foundation and the U.S. Department of Agriculture.

According to an announcement last week, Washington State University will spearhead the new USDA-NIFA AI Institute for Agricultural AI for Transforming Workforce and Decision Support, or AgAID Institute, to develop machine learning methods to help farmers navigate challenges associated with climate change, such as drastic weather changes and water supply management.

Ananth Kalyanaraman, a WSU computer science professor and lead principal investigator for AgAID Institute, said he'll head a coalition of agriculture workers and researchers from nine other institutions to design AI to guide water allocation and extreme weather response efforts.

According to Kalyanaraman, water and resource management will be crucial as climate change threatens water supplies across the globe. He said AI could help farmers make the most out of water availability, noting that agriculture accounts for about 80 percent of water usage in the U.S.

“AI has had a demonstrated potential to address socially relevant problems like agriculture,” he said. “In other words, ‘How can we harness data to make informed decisions and provide significant improvement in agricultural productivity?’”

For example, an orchard can use water or fans to ward off the effects of an impending frost, depending on the extent of the frost being forecasted. Kalyanaraman said AI can help farmers decide cost-effective, resource-efficient approaches to such scenarios, with specific conditions in mind.

He said part of the overall goal is to “partner human knowledge with AI tools in a way that amplifies the end outcomes."

“We will be developing new AI methods to help with a more effective and efficient use of data,” he said. “Using forecasting and seeing how a crop is responding, we can come up with actions to recommend for a farmer to pick.”

Kalyanaraman said the university has already outfitted "smart farms" in Washington with sensors on drones that collect multispectral and thermal imagery data in an effort to gauge changes in weather. From there, the AI can help guide decisions related to temperature changes and water availability.

“A lot of these farms are already instrumented and data is being collected, but not much is being done with it. That’s the gap we are planning to close," he said.

“I think we have got to a stage where AI can be reliably used to bridge the gap between models and data,” he added. “That is critical because it’s going to help us improve the models by making them more robust, and provide better bounds of uncertainty that can be extremely useful to the farmer to make decisions.”

While researchers led by WSU focus on agricultural tech, another NSF-backed venture, the AI Institute for Dynamic Systems, will be led by the University of Washington and focus on creating an AI to respond quickly to unpredictable and complex engineering problems.

According to a news release, this institute will “integrate physics-based models with AI and machine-learning approaches to develop data-enabled efficient and explainable solutions for challenges across science and engineering.”

J. Nathan Kutz, a professor of applied mathematics at the University of Washington, said the real-time machine learning needed to create autonomous technologies, such as self-driving vehicles or robotics for complex mass production, remains a key challenge for AI researchers.

“When you’re in a self-driving car, you’re always encountering new environments and new worlds, and so the AI is forced to have to extrapolate. Or in other words, ‘I haven’t seen anything like this before, so how do I make good decisions in this environment?’” he said. “That’s still a big challenge, and it’s actually what a lot of the focus of our institute is going to be developing algorithms around.”

Kutz said there are still many limitations to what AI can and can’t do when it comes to decision-making, which impedes the adoption of reliable autonomous technology for various functions.

“Humans can adapt to changing conditions, whereas AI has a very hard time adapting when it’s never collected data on something it’s never seen and has to collect data there to build a model on,” he said.

The two institutions are among 11 National Artificial Intelligence Research Institutes that received a total of $220 million in grants announced last week by the NSF. Each will receive $20 million in grant funding over five years for their research, set to begin in the fall.

Kalyanaraman said he's looking forward to the possibility of collaborating with other institutes in the years to come.
Brandon Paykamian is a staff writer for Government Technology. He has a bachelor's degree in journalism from East Tennessee State University and years of experience as a multimedia reporter, mainly focusing on public education and higher ed.