Drones and sensors join the fight against wildfires
Wildfires have now burned more than 4.6 million acres across 10 western states and are responsible for at least 33 deaths. Oregon Gov. Kate Brown said the wildfires could lead to the greatest loss of property and human lives in her state’s history. Unfortunately, the frequency of catastrophic fires is expected to escalate across western regions of the U.S. due to increased summer temperatures and earlier spring snow melts.
Wildfires are gruelingly difficult to fight, and their behavior and spread are notoriously hard to predict. Although not trained to fight fires, Mechanical and Aerospace Engineering Associate Professor Mrinal Kumar and his team are developing technology to help wildland firefighters determine where and how the flames will move.
Kumar leads the Laboratory for Autonomy in Data-Driven and Complex Systems within The Ohio State University Aerospace Research Center. Here they focus on the quantification of uncertainty in complex systems. Severe nonlinearity, high-dimensionality and unorthodox uncertainty make a wildfire the quintessential complex system. The lab also specializes in the guidance, navigation and control of unmanned aerial vehicles (UAVs). The combination of the two areas of expertise has resulted in a multi-dimensional approach to wildfire modeling and prediction.
“There is a growing need for rapid, real-time modeling that can precisely represent current and future wildfire areal coverage to assist responders,” said Kumar.
His team’s approach combines forecasting—based on probabilistic environmental conditions—with in-field temperature sensors and mobile UAVs with vision and infrared cameras. The aerial drones carry sensors to the heart of the fire and transmit live images to mission control, where the data is integrated with forecasting data and other sensor data in what Kumar calls “a novel evidential information fusion paradigm.”
“This fused information improves our knowledge about the fire, as well as helps inform where the UAVs should travel next to gather information,” Kumar explained. In addition to the algorithms for flight planning and path following, his team also builds the actual UAV platforms that fly over the blazes.
Currently the most widely used wildfire modeling application is FARSITE, which computes fire growth and behavior for long time periods under heterogeneous conditions of terrain, fuels, and weather. The U.S. Forest Service, National Park Service, and other federal and state land management agencies rely on it during firefighting.
Ohio State’s approach aims to augment and assist forecasting tools like FARSITE, which does not incorporate online sensor data. Kumar said that sensors are an integral part of this approach, which distinguishes “uncertainty” from “ambiguity” so that conflicting data does not lead to complete loss of information.
“Our work is to develop a complete information fusion and mission planning platform, of which FARSITE can become an important ingredient,” Kumar said. In 2019, he teamed up with former grad student Alex Soderlund and former post-doctoral researcher Chao Yang to earn the AIAA Sensor System and Information Fusion Best Paper Award based on the lab’s research.
In collaboration with Associate Professor of Forest Ecosystem Analysis and Management Roger Williams, Kumar’s team has conducted tests of their technology in prescribed burns conducted by the Ohio Department of Natural Resources (ODNR) in prairies near Ohio State’s Marion campus.
“We have demonstrated that the evidence-based information fusion vastly outperforms standard forecasting where no sensor data is incorporated, as well as other fusion paradigms in which it is assumed that all sensor data is good,” Kumar said.
Planned tests in the forests of Coshocton, Ohio, were canceled due to weather, and the pandemic has prevented rescheduling. Kumar’s team will continue to work with ODNR and acquire further research grants for building operational systems, with hopes to expand westward over the next couple of years.
by College of Engineering Communications