Drones on a mission. Professor Kyriakos Vamvoudakis and his research team are developing a new class of drones with life saving abilities. 

Conventional aviation operations used for today’s wildfire management put the life of pilots, fire-fighters, and response teams at risk of injury or death. 

The National Science Foundation has awarded a $1.2 million grant to the Daniel Guggenheim School of Aerospace Engineering, U.S. Forest Service, Kaibab National Forest, and the Arizona Department of Forestry and Fire Management in a collaborative effort to perform multiple field tests to combat wildfires.  

Kyriakos G. Vamvoudakis

Professor Kyriakos G. Vamvoudakis and his team will research and develop frameworks for unpiloted aerial vehicles (UAVs) that offer safer detection, prevention, mitigation, and collaboration for wildfire management. 

As the principal investigator, Vamvoudakis plans to use data shared between UAVs and ground vehicles to develop data-driven distributed methods to find optimal ways for fire management and evacuation. For instance, officials can use this technology to safely monitor the fire frontline and deploy teams based on their available resources and distance to the impacted area.

“The high numbers of naturally occurring and prescribed fires in the southwest provides abundant opportunities for our team to work together with stakeholders to test and improve UAV operations,” said Vamvoudakis. 

The UAVs will use reinforcement learning, an area of machine learning, and extremum seeking for real-time optimization to provide base-line information that can be expanded in future research. 

The team will also work to find the safest evacuation path by autonomous UAVs to help guide ground vehicles, residents (potential tourists), and firefighters using a novel bounded rational game-theoretic framework that predicts the future states on an unknown and dynamic environment. 

“Using bounded rationality theory allows us to measure the unpredictable nature of wildfires and provide adaptive countermeasures – like identify the fastest and safest evacuation roads for firefighters and fire-trucks in highly dynamic and uncertain dangerous zones – given limited information and time constraints,” explained Vamvoudakis. 

The results of this project will have a broad impact on several applications, such as target identification and tracking, search-and-rescue missions, and disaster management. Eventually, it will be used as a model for wildfire management in other parts of the country and the world.