ARLINGTON, Va. - Inside the bustling confines of an operations center, National Guard members quickly assemble information from multiple sources, piecing together the “big picture” from ground and air personnel as they fight a devastating wildfire threatening lives and property.
As the day progresses, their report — meant to be a timely representation of the situation — makes its way up the ladder, subjected to revisions and interpretations at every level.
By the time the information reaches key decision makers, however, the realities on the ground have long since evolved: homes destroyed, wildlife scattered, vegetation wasted, and most importantly, lives lost — a devastating situation exacerbated by delays in communication.
For Michael Wisniewski, chief data officer with the Army National Guard’s communications and computers directorate, such a scenario should never become a reality in the fast-paced world of natural disaster response.
“Back in the day, we’d fly over a fire or disaster site, then come back, look at a map, make some marks on it, and then go out again and come back,” said Wisniewski. “I jokingly say it’s also like using World War II-level communications: Somebody delivers a sheet to somebody else, and then they read it to the commander.”
However, those holdover processes are becoming a thing of the past as the National Guard begins to embrace artificial intelligence to help it see and understand situations quickly and employ resources faster to help save lives and property.
And significantly reducing that lag in communication via AI is the aim of Project Theia — an example of how the Guard is committed to modernization, said Army Gen. Daniel R. Hokanson, chief of the National Guard Bureau.
“It is widely known the National Guard has been at the forefront of domestic response efforts — especially during the past few decades,” he said. “And this technology, simply put, is going to take us to the next level in not only how we coordinate with local agencies, but the speed in which we can help save lives.”
Named after the Greek goddess of sight, Theia centralizes video data and then applies AI solutions to increase situational awareness among those responding to natural or human-caused disasters.
“We are trying to get all the imagery that exists in the Guard and other military assets, dump it into one platform and distribute it accordingly,” said Wisniewski.
The technical aspects of the project begin with taking in video streams from military air assets along with imagery and other information from ground assets.
“But whether it’s a [UH-72A] Lakota military helicopter, MQ-9 Reaper [unmanned aerial system] or a civilian partner’s assets, the process of getting that footage is entirely different for each platform,” said Wisniewski.
This presented the challenge, he said, of rapidly creating a common operating picture for response forces.
“So we built a system that said, ‘Look, we’ll handle the integration of all the systems, and then standardize the metadata behind the full-motion imagery coming in to the point where I can run AI algorithms on it,’” Wisniewski said.
The result is a web-based tool that provides clearer depictions of disaster areas, said Air Force Capt. Douglas Witherspoon, with the California Air National Guard’s 163rd Attack Wing.
“Now, artificial intelligence can take data from a video and create an overlay and then allow the aircraft’s video program to connect to it,” he said. An overlay involves using Theia’s software to render markings on top of a digital map of the disaster site.
Because the overlay is constantly updating the imagery, Witherspoon added, analysts can better track fires. Whereas before, he said, those same analysts manually created overlays on a printed map or applied them to a digital one — both time-consuming tasks.
“The technology we have now is more advanced and fully automated,” Witherspoon said. “Using artificial intelligence to visually augment live video feeds with highly accurate overlays of roads, key landmarks, and other mission-critical data allows analysts to make time-sensitive decisions.”
But first-line responders and ground operators benefit too, said Army Col. Daniel Bowles, vice director for the National Guard Bureau’s communications and computers directorate.
“You still have radio and satellite communications,” he said. But one of Theia’s key features is “getting real-time information to the very person that’s on the ground, as fast as we can make it and [enhancing] decision-making capabilities.”
The project also was an exercise in collaborating with academia. Through Hacking For Defense, a Defense Department-sponsored university course, students work with defense and intelligence communities to rapidly address the nation’s emerging threats and security challenges.
“In the National Guard, we have a warrior’s mindset when it comes to fixing a problem, which is not a bad thing,” said Bowles. “But then we also have the mindset of a Citizen-Soldier or Airman — meaning that we come from industry, we come from the workforce, we come from academia, and we can provide multiple viewpoints.”
Tapping into the talent of young minds in academia, he added, is a natural fit for the Guard.
“There are so many different solutions that are out there and when we look outside the box, we can leverage those different things,” Bowles said.
Yet with the technological wonders of AI, Wisniewski said the project highlighted another accomplishment.
“The real takeaway is the transformative approach to standardizing all these different formats of videos that are coming in — so we can apply AI solutions,” he said.
With nearly 400 domestic response operations under its belt last year, Bowles said the Guard will continue to strategize how it can integrate AI to better support federal and local officials during disasters.
“We’re not just using new technology just to use it,” said Bowles. “There’s got to be a reason — an understanding on how it makes a process better.”
Project officials expect to support the California National Guard during the upcoming wildfire season. They think the system will be fully operational by the end of the year.