Drones can help us with wildfire detection and suppression.
Matthew Spencer, associate professor of engineering at Harvey Mudd College, takes a bird’s eye view.
Matthew Spencer (Member, IEEE) received B.S. and M.Eng. degrees from the Massachusetts Institute of Technology in 2007 and 2008, respectively, and a Ph.D. degree from the University of California, Berkeley, in 2015. He is currently an Associate Professor of engineering with the Harvey Mudd College. His engineering interests include high-speed communication circuits, instrument interfaces to “weird physics,” microelectromechanical systems, and engineering education. His work has won a variety of awards, most recently the best-in-division paper from the electrical and computer engineering division of the American Society for Engineering Education annual conference in 2022.
Drones for Wildfire Detection and Suppression
Wildfires have devastating effects on lives and property. For example, the recent Eaton fire in California resulted in $250B of property damage and took 17 lives. The intensity and frequency of wildfires are expected to increase as climate change progresses, so technological solutions to help suppress wildfires are crucial.
Research shows that early responses to wildfires can reduce the damage they cause; some estimates suggest that a 15-minute reduction in response time can prevent up to $8B in damages. One way to reduce the response time to fires is to detect them very early, when fires can be as small as a few feet across. This is challenging because wildfire often starts in undeveloped wilderness, so early warning systems must observe hundreds or thousands of square miles for a tiny fire. Even worse, typical indicators of wildfire—like smoke plumes—are only visible long after a fire has started.
Fortunately, a combination of drone technology, sensor technology and AI can help spot tiny fires. I am collaborating with a company called ThinkCircuits to build a network of drones that can automatically detect and suppress wildfires. We are working on two innovations that will enable this drone network. First, we’ve created a specialized sensor that can detect small fires at long distances; we’ve had success spotting a 2-inch flame more than six miles away. Second, we’re using very efficient AI models that will allow fire suppression drones to decide how to use their limited payload of water on rapidly spreading fires. The drones need to make decisions quickly, so they can’t wait for human input and instead rely on these models, which must run on the drones’ small onboard computers.
