Re-imagining the Edge: How Agencies Can Tap Edge AI to Achieve Mission Outcomes

Wildfires continue to devastate communities across America. In 2021 alone, 40,000 fires ravaged the nation, and that number is expected to increase by year's end. What’s more, the U.S. Forest Service has sent roughly 21,000 federal firefighters to put out the fires, twice the number of responders needed to mitigate fires at this time last year. As new wildfires emerge, officials fear they may not have the resources needed to solve this widespread problem.

Edge artificial intelligence could be the solution they need.

“When we think about applications of edge AI in the modern world, we're always going to be thinking about how we respond to disasters,” says Rob Risany, director of edge AI solutions architecture at Intel Corp. “Let’s say I’m managing a forest fire. If I deploy an edge-based inference system out on a mobile device and drop it out in the middle of a forest, I can use sensor data to sense the temperature, [how fast the] smoke is moving or how the wind is changing. And that can advise my firefighters in a much stronger sense what's going to be happening next.”

Edge AI, Risany adds, can also assist first responders in other critical situations, like a collapsing building or an active-shooter incident.

“When you’re dealing with any type of situation where people are in harm's way, we have to recognize that edge AI is also pushing the network closer to where the people are,” he explains. “So, instead of having to count on the fact that we have large backhauls back to the cloud, or back to some data center, we can now actually deploy mobile edge compute right out to where it's needed.”

Moreover, edge AI empowers first responders to make timely decisions quickly and accurately so they can ensure citizens are safe from danger, by enabling communication among the teams on the ground.

“If there's no communication available in that area, we now have an opportunity with 5G to stand up a small private network that is secure,” says Mike Broadwater, vice president of GAI’s technology services group. “That can process data both as a data center and as an edge AI base.”

Consider, for instance, a hurricane response team tasked with finding and identifying survivors. Thanks to edge AI, these first responders can receive real-time information about, for example, a tree blocking the road or downed power lines in a specific area. With these insights, responders can refocus their efforts on areas they can quickly access.

“And then, as the other teams are taking care of those trees and power lines, they can start responding in those areas,” Broadwater notes. “That type of a coordinated response really is helping that frontline first responder to get out and do what they're supposed to: save lives and help support the people in a disaster.”

Disaster response and management might be the most timely opportunity for agencies to embed edge AI into their operations, but it isn’t the only one. Airport and transportation security could also reap the benefits of this technology.

“Even before [COVID-19] hit, we saw several efforts to streamline our airports, we saw the ways that we were incorporating new types of analytics to find and examine X-rays and provide coaching to [Transportation Security Administration] specialists who were new on the job and didn't know exactly what to look for,” Risany says.

In short, edge AI empowers these critical government employees to do their jobs better.

At agencies like TSA, the Federal Aviation Administration and Customs and Borders Protection, maintaining the workforce is a key challenge, Broadwater says.

“For example, do we have enough people to be at every port of entry at every TSA gate? Do we have enough people that understand not only what the technology is doing, but even how to apply that in their day-to-day tasks?” he says.

Once agencies understand what edge AI can do for them, the question then becomes, how to get started? As with any emerging technology, bringing security into the process early on will be critical to success.

“Security … can no longer be a bolt-on type of a scenario,” says Chris Thomas, lead systems architect at Dell Technologies. “It's not your car, where you add features and functions, change the color or [get a] new paint job. It has to be something that you think about from the beginning.”

And, because this technology is still very new, education is also key to successful edge AI adoption.

For instance, “do we even understand what we're asking to be deployed?” Broadwater asks. “Do we really understand what it can do for us and how it can help with our mission? If a director says, ‘I need to deploy edge analytics to run natural language processing,’ do the folks underneath that director actually understand what that means? Does the director actually understand what that means? Making sure we're all speaking from that same data dictionary is ... key to our government being successful in this space.”

Indeed, effective adoption of edge AI is only possible with the right people in place to make innovation happen. According to Broadwater, organizations need staff who can deploy, understand and maintain algorithms.

He also recommends agencies begin with a proof of concept that allows them to develop a plan and execute against it in a controlled environment.

“This gives you the best readiness assessment that you could have done,” he says. “You will understand where your weaknesses are, how you're going to be able to [get] things into the field, how you're going to get that data back, what you're going to do with the information that you have.”

The more agencies experiment with edge AI concepts, the sooner they will start to see results.

“The value of edge AI can't be [over]stated, because [it] enables a modernization approach that allows an agency to skip many steps, which makes for a really great return on investment and, importantly, a great impact for the citizen,” Risany says.