IE 11 Not Supported

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

Research Uses AI, Machine Learning to Monitor Soil Moisture

A project from the University of Maine monitors soil moisture using AI that over time learns to make a sensor network as efficient as possible, creating a low-energy way to track climate change impacts on natural environments.

Ali Abedi (center) and the wireless soil moisture sensor with graduate student Kenneth Bundy (left) and local high school teacher Ed Lindsey (right).
University of Maine
Monitoring soil moisture is an important component of both forest management and agriculture, but existing systems need a ton of energy to work well and can be prohibitively expensive for researchers and farmers. But keeping track of soil moisture is essential to keeping ecosystems strong, especially as drought conditions worsen in some parts of the world.

With funding from the National Science Foundation, a new project from the University of Maine’s Wireless Sensor Networks (WiSe-Net) lab uses AI and machine learning to make soil monitoring more efficient. Working with researchers from the University of New Hampshire and University of Vermont, WiSe-Net developed a network of AI-powered sensors that learn over time how to better use energy while monitoring soil moisture and processing the resultant data. That means it will adjust as needed to take advantage of available wireless network resources.

“AI can learn from the environment, predict the wireless link quality and incoming solar energy to efficiently use limited energy, and make a robust low-cost network run longer and more reliably,” said Ali Abedi, professor of electrical and computer engineering at the University of Maine and principal investigator on the project.

The technology can also be used for other types of sensors, using the same AI methods to measure things like snow depth.

“Real-time monitoring of different variables requires different sampling rates and power levels,” Abedi said. “An AI agent can learn these and adjust the data collection and transmission frequency accordingly rather than sampling and sending every single data point, which is not as efficient.”
Lauren Kinkade is the managing editor for Government Technology magazine. She has a degree in English from the University of California, Berkeley, and more than 15 years’ experience in book and magazine publishing.