Artificial intelligence gets real

Technology being developed in a nondescript office building in Reston could change how Army soldiers train for and operate in combat thousands of miles away.

The Army’s Integrated Visual Augmentation System (IVAS) runs on a pair of Microsoft goggles and links to a microsized drone that flies autonomously and collects video analyzed in real time by artificial intelligence algorithms trained to identify threats, like an enemy combatant with an assault rifle coming around a corner, or a vehicle of interest. Detections are sent to a heads-up display within the goggles and are shared across a squad.

“It can all be done at the tactical edge out on the battlefield, using new-edge computing technologies, which basically puts the power of a supercomputer in the soldiers’ hands,” says Rob Albritton, a vice president at Reston-based Octo who heads up the AI Center of Excellence at the federal contractor’s oLabs tech accelerator. Octo has been working on developing AI technology for IVAS since 2020 and is currently working with about 20 government agencies on a variety of other AI projects.

In 2011, IBM’s Watson supercomputer wowed the world, demonstrating the power of artificial intelligence by beating “Jeopardy!” champ Ken Jennings, who still holds the television quiz show’s record for the longest streak of victories. Bowing to defeat, in his response to the match’s final question, Jennings added a “Simpsons”-inspired quip: “I for one welcome our new computer overlords.”

While that sort of doomsday scenario hasn’t occurred, artificial intelligence has matured into a routine — and often unseen — part of our daily lives, moving in recent years from the realm of science fiction to the mundane. From mapping apps on smartphones to technology that helps secure the United States’ borders, AI is performing everything from rote business tasks like scanning health records to helping the Navy search for underwater mines from unmanned surface ships.

And several Virginia-based companies, running the gamut from small tech startups with a few clients to massive Fortune 500 federal contractors, are at the leading edge of AI work.

“AI will enable how … systems navigate, perform their mission, sense the environment and communicate strategic information to stakeholders, which are both people and other systems,” says Duane Fotheringham, president of unmanned systems for Huntington Ingalls Industries Inc.’s mission technologies division. “On the data side, humans alone cannot manage the volumes of time-sensitive data that will be collected and shared by autonomous systems.”

From small to large

Using its cloud-based EarthAI platform and free data from the European Space Agency, Charlottesville-based Astraea Inc. mapped out every rooftop with a white membrane — a waterproof, reflective covering preferred by the solar industry for installation — and matched each with an address and owner or landlord across six counties in Virginia after Astraea was approached by a solar company client for help pinpointing customers.

Though it took Astraea time to build its algorithm, it took EarthAI just five hours to complete the scan. A job that might have taken nearly two dozen employees a year to research was completed in a fraction of the time with up-to-date data, using artificial intelligence, says Astraea co-founder and CEO Brendan Richardson.

Founded in 2016, Astraea has a few dozen enterprise customers, including The Nature Conservancy, which uses EarthAI across multiple areas, including land preservation and carbon sequestration, tracking forest changes in remote areas from logging, road construction, clear-cutting, tree disease and replanting. Astraea expects to double in size to about two dozen employees by the end of the year, with revenue in the “low millions,” Richardson says.

At the other end of the spectrum, Chantilly-based federal contractor Peraton Inc. has a portfolio of about $150 million in contract work related to employing advanced data analytics and AI for customers such as the Department of Defense and the Centers for Medicare and Medicaid Services, says Mark Adams, Peraton’s vice president of technology and engineering. Because of the ubiquity of AI, however, Adams says that $150 million represents only a portion of the company’s AI projects.

“There are a lot of jobs where we’re doing large-scale, enterprise IT and even mission-critical, large-scale cyber and intelligence networks where we’re actually building the underlying infrastructure and secure data management system” using AI, Adams says. And the company is looking into other ways of expanding the use of AI to support its customers.

Adams and others say they expect the use of AI to continue growing, and Virginia companies can expect to grow with it.

In June, Reston-based federal contractor NCI Information Systems Inc. rebranded as Empower AI Inc., taking the name of the AI platform it launched in 2020. Depending on the level of automation desired and the complexity of the task, the platform is adaptable and can be customized to a client’s needs, says Empower AI Chief Technology Officer Allen Badeau.

During a nine-month period this year, Empower AI’s automations saved one federal client, which he declined to name, as much as 13 man-years worth of work by completing up to 7,000 automations a day, including verifying information found in health records.

“When you’ve got health records potentially that are hundreds and, in some cases, even thousands of pages long, to move that kind of information between those systems takes a long time,” Badeau says. “When you can automate that process and automate some of the validation around that, it’s a significant time-saver from a human perspective.”

More than 75% of Empower AI’s revenue now comes from its AI platform, which the company says accounts for more than $1.5 billion worth of work. Empower AI has more than 20 federal customers, a figure Badeau expects will double in the next year.

‘Never get tired, bored or distracted’

Recent developments in the AI field helped pave the way for Falls Church-based General Dynamics Information Technologies Inc., a subsidiary of Reston-based Fortune 500 aerospace and defense contractor General Dynamics Corp., to develop an algorithm that can classify skin lesions, helping medical personnel diagnose skin cancer.

GDIT’s Skin Lesion Classifier prototype won third place and $10,000 in a U.S. Department of Veterans Affairs competition that ran from 2020 to 2021. GDIT has pitched the idea to the VA as a pilot project.

“The ability to take a picture of a skin lesion and quickly help an oncologist decide if it’s a particular kind of a disease, and then whether it’s malignant or benign, really speeds the workflow so that veterans are seen faster and their appointments are escalated if they have dread diseases,” says Dave Vennergrund, GDIT’s vice president of AI and data insights.

Streamlining systems for veterans is also an area that Tifani O’Brien is addressing in her work leading an AI and machine-learning accelerator at Reston-based federal contractor Leidos Holdings Inc. 

As part of two Veterans Benefits Administration contracts awarded this year to a Leidos subsidiary to provide medical disability exam services for veterans and separating service members, Leidos is incorporating AI to screen intake questionnaires for disability claims, using the tech to route patients to the correct physicians with greater speed and accuracy. For the other contract, Leidos is using AI to screen thousands of pages of handwritten medical records, freeing up doctors to focus on the relevant pages they need to make health care decisions for their patients.

The contracts, both of which have six-month base periods and six one-year options, are worth up to $1.7 billion if all options are executed.

“We’re making it faster and more cost-effective to make decisions on what’s the next step in a veteran’s access to services and what they need to do next,” says O’Brien, a Leidos vice president. “Hopefully, that frees up more time to actually spend talking to the veteran and servicing the veteran and interacting with them, as opposed to going through the documentation and bureaucracy and deciding what to do next.”

While Leidos’ medical AI work is more behind the scenes, the company’s AI tech is showcased in a considerably more visible way at airports across the nation. If you’ve gone through an airport security checkpoint recently, chances are you’ve encountered Leidos’ screening machines, including ProVision2, a body scanner, and ClearScan, a luggage scanner. Both products use AI to determine quickly if a passenger or their bags contain prohibited items that checkpoint workers might not be able to identify in a cluttered space.

The AI can pick out objects based on characteristics that a human might not see, explains Ron Keesing, Leidos’ senior vice president for technology integration. “Also, computer-vision algorithms never get tired, bored or distracted,” Keesing says. “In this way, Leidos AI works as a partner to humans, helping them do their jobs more effectively and efficiently.”

Unmanned future

When it comes to AI development, the Pentagon is a focal point for many of Virginia’s largest federal contractors and tech companies. In the race to keep up with adversaries like China and Russia, the Pentagon in 2018 established the Joint AI Center — now known as the Chief Digital and Artificial Intelligence Office — to accelerate delivery of AI national security operations.

In 2020, McLean-based Booz Allen Hamilton Inc. received an $800 million, five-year task order to integrate and develop AI for the warfighter. Booz Allen’s work on that contract, called the Alliant 2 Joint Warfighter Task Order, runs the gamut from data labeling, management and conditioning to AI product development and incorporating the tech into new and existing programs and systems across the Department of Defense.

Booz Allen is the prime contractor on more than 150 AI projects for the federal government, says Eric Syphard, a vice president in Booz Allen’s artificial intelligence engineering practice.

The Fortune 500 company also started a venture capital fund this year to invest in AI companies, helping sharpen its focus on emerging capabilities. In April, Booz Allen announced an investment in San Francisco-based Reveal Technology Inc., which developed a platform, Farsight, for military special operators and squads operating in austere environments. The platform relies on autonomous drones to develop maps, line-of-sight analysis and route planning without the need for network connectivity.

“Imagine being a troop deployed to an area that you’re brand-new in that geography, you want to get a handle on some potential threats, the infrastructure and roads and you want AI to help you plan and recommend which path to take,” Syphard says, adding that the capability is similar to a civilian using a GPS traffic app to determine the fastest route to take during rush hour.

The individual military branches, too, have noted the importance of AI.

In April, HII’s  mission technologies division, based in McLean, announced the launch of Odyssey, an open-architecture platform that can turn any ship or vehicle on land or in the air into an intelligent, autonomous system. While it’s got a range of possibilities, Odyssey can use sensors to autonomously navigate, and it can be used to automate mechanical or electrical systems and diagnose problems when issues may arise.

“Sometimes we talk about the ‘AI captain’ steering the ship, but there also has to be an ‘AI engineer’ that’s controlling all of the engine-room operation,” says Fotheringham. “Then we also have to make sure that the systems are both reliable and resilient because we all know that things are going to break.”

The technology is already being incorporated into an unmanned, surface-traveling maritime vessel that will be used by the Navy to hunt and sweep for mines. Louisiana-based Bollinger Shipyards in April received a contract worth up to as much as $122 million to build as many as 30 unmanned vessels that will rely on Odyssey during the next five years. HII is also part of a team contracted for work on initial concept design for the Navy’s large, unmanned surface-ship vehicle program.

Fotheringham notes that the Navy has projected that up to 30% of its force structure could be unmanned in the future, and AI will be “the fundamental enabler” of that.

In September, the Army accepted delivery of its first 5,000 IVAS goggles, Bloomberg reported. The service spent $373 million on that order and could pay out as much as $21.9 billion on as many as 121,000 of the devices during the next decade.

While an Octo spokesperson declined to comment on whether that order incorporates the company’s technology, Albritton says the program has far-reaching potential.

“If this program is successful, there’s no doubt in my mind that we’re gonna save lives,” Albritton says. “And being able to bring more soldiers home to their families is really why we do what we do.”