U.S. Markets open in 33 mins

'A.I., Captain': The Robotic Navy Ship of the Future

Aaron Pressman
1 / 7

All Aboard the Navy's First A.I. Warship

All Aboard the Navy's First A.I. Warship

The swells in the middle of the North Pacific were reaching nine feet when one of two engines on the diesel-powered U.S. naval ship called Sea Hunter shut down. About 1,500 nautical miles from its home base in San Diego, the 132-foot-long craft, which had been cruising at 10 knots, couldn’t send a member of its crew to check out the problem—because it didn’t have a crew.

Sea Hunter’s sleek, spiderlike silhouette, with a narrow hull and two outriggers, is a prototype of what could be a new class of autonomous warships for the U.S. Navy. Its artificial intelligence–based controls and navigation system, designed by Leidos Holdings, a defense contractor based in Reston, Va., were seven years in the making. And this maiden voyage—a more than 4,000-mile roundtrip to the giant Pearl Harbor naval station—was its first major proof of concept.

Nothing like this had ever been attempted before. And while the A.I. systems that keep the ship on course and help it avoid collisions with other vessels were working exactly as advertised, a glitch in its mechanical systems threatened to scuttle the trip—a reminder to tech geeks that no matter how advanced the technology, mundane mechanical problems can bring a project down.

A group of 14 support staff in a trailing escort ship sprang into action. Keith Crabtree, a systems engineer with Leidos, and other staff jumped into a rigid inflatable boat and zipped over to Sea Hunter. Crabtree, who had helped put the ship through its paces in the calmer waters of San Diego Bay, says he wasn’t worried about the swells as he rode across the waves to Sea Hunter. The triple-hulled design of the prototype, inspired by the Polynesian waka canoe, offered a more stable perch than the bouncing journey aboard the escort ship.

“We were in for a smoother ride than what we had been enduring,” Crabtree recalls. A simple software fix corrected the problem, and after docking at Pearl Harbor, Sea Hunter completed the 10-day return trip without incident.

Sea Hunter, it bears noting, is the first autonomous ship to make an ocean crossing and, remarkably, the first Navy ship designed from scratch by Leidos.

Little known outside government contracting circles, Leidos, then dubbed Science Applications International Corp. (SAIC), was founded 50 years ago by Robert Beyster, a brilliant and entrepreneurial physicist who had worked on the hydrogen bomb at the Los Alamos National Laboratory. An avid sailor and a friend of yacht-racing captain Dennis Conner, Beyster tasked SAIC to develop software to model improved hull designs after Conner’s squad lost the America’s Cup to an Australian team led by Alan Bond in 1983—the first American loss in the race’s 132-year history. Connor regained the Cup the following year.

Roboship at rest: The Sea Hunter docked in San Diego, its home-base harbor. | Photograph by Spencer Lowell

That expertise came in handy on future projects with the Navy but didn’t publicly reemerge until 2012, when a $59 million contract win to develop an autonomous ship put the software front and center once again. For Sea Hunter, the company also drew on expertise gained from many loosely related projects, including developing underwater sensors for the Navy, performing coastline surveys for the National Oceanic and Atmospheric Administration, and conducting A.I. work to process satellite imagery.

That’s exactly the kind of eclectic mix of tech-savvy competencies that have underpinned Leidos’s five-decade existence as an under-the-radar but important Pentagon contractor. With $10.2 billion in revenues last year, the company is ranked 311 on the Fortune 500 for 2019—its third straight appearance on this list.

While defense and intelligence work generates nearly half of revenues, Leidos has its hands in virtually every aspect of the federal government’s technological and logistical efforts, including running the Frederick National Laboratory for Cancer Research, designing a microwave system for military vehicles to detect IEDs, and building a digital medical records system for the Defense Department. Analysts expect the firm’s revenue to rise 5% this year, to $10.7 billion, with earnings climbing 8%, to $627 million.

For now, Sea Hunter isn’t even a blip on Wall Street’s radar screen, but it could become a big growth driver if Leidos wins a major role in the Navy’s upcoming plans to add a dozen or more autonomous ships.

That’s a big if. Just because Leidos designed the prototype for Sea Hunter doesn’t guarantee it a role in the multibillion-dollar contracts to come. In the ruthless world of defense contracting, lawsuits and protests are common; Leidos was bumped from one $2 billion bidding battle for a Justice Department IT contract in 2018 when a competitor complained that a pricing spreadsheet had some blank cells. “What keeps me up at night is someone else claiming they can do better,” says Rus Cook, Sea Hunter’s senior program manager. “That would just be a huge waste of the taxpayers’ money.”

The A.I. software it has developed so far could give Leidos a big leg up. No other company has publicly demonstrated anything close. “They’ve got the archetype out there in the water, doing its thing on the open ocean,” says Bryan McGrath, a retired 21-year Navy veteran who is now deputy director of the Center for American Seapower at the Hudson Institute. “It’s really exciting for the future.”

Long a well-regarded government contractor, Leidos predecessor SAIC suffered from almost 10 years of problems after the 2004 ouster of founder and CEO Beyster, who opposed taking the company public. The Obama-era defense budget cuts hammered the company’s revenue growth and contributed to the first-ever operating loss in Leidos history. And most damaging, a massive scandal involving a New York City payroll project landed two executives in jail and resulted in fines and restitution costs totaling more than $500 million. At the same time, the federal government tightened its conflict-of-interest rules, prompting big contractors like Northrop Grumman and Lockheed Martin to spin off their services divisions.

So in 2012, the company moved to shrink itself by splitting in two. A technical services unit, which performed tasks like upgrading military vehicles and assembling flight simulators, was spun off under the SAIC name. The larger information technology and sciences unit went forward as Leidos. The name was created by lopping off the front and the back of the word kaleidoscope.

To run the new operation, Leidos hired Roger Krone, now 62, an aerospace engineering graduate from Georgia Tech who holds an MBA from Harvard. Before joining Leidos in 2014, he served in senior positions of finance and project management at Boeing, McDonnell Douglas, and General Dynamics.

Computer-nerd CEO Krone, an engineer by training, began his career writing programs on IBM punch cards. | Photograph by Scott Suchman

In a well-tailored navy suit with an on-brand purple tie, the Leidos corporate color, and a distinguished shock of silver hair, Krone could fit in easily on Capitol Hill among the senators and lobbyists at any hearing. But the CEO is a computer nerd at heart, recounting stories of his earliest programming days growing up near Cincinnati with a TRS-80 home computer that he upgraded himself and writing programs on punch cards for an IBM 360 mainframe at nearby Xavier University. The part-time punch-card job, which paid $7 an hour, came after his programming teacher caught him working the ovens at Pizza Bob’s in Cincinnati and challenged him to put his skills to better use. But finding the bugs and rewriting code didn’t suggest a satisfying career path, so he went into aerospace engineering, eventually helping design airplanes, helicopters, and spacecraft.

A year after taking over as CEO, a bargain of sorts fell into Krone’s lap. Defense giant Lockheed Martin had spent $9 billion to acquire Sikorsky Aircraft and decided to raise some money by selling its IT businesses, a jumble of units with total sales of about $5 billion. For $4.6 billion, paid mostly in stock, Krone gobbled up businesses whose work included designing a next-generation air-traffic control system, billions of dollars of IT programs for the Social Security Administration, and a host of military projects.

The deal, which nearly doubled Leidos’s revenue, closed in the summer of 2016, coinciding almost perfectly with the arrival of the Trump administration and major increases in defense spending to combat the growing military presence of China and smaller threats from North Korea and Russia.

Having successfully integrated the acquisition, Leidos is on the upswing; revenue last year was twice the $5.1 billion the company booked for the 12 months before the merger. And in the first quarter of this year, the Leidos backlog of business grew to a record $21.5 billion, aided by a $3 billion contract to run NASA’s IT network for up to 10 years. Leidos’s share price, at $75 recently, has returned 207% since Krone took over in July 2014. That compares with a 60% rise in the S&P 500 index and a 106% gain for the Dow Jones U.S. Select Aerospace and Defense index.

In addition to defense and intelligence work, Leidos is deeply involved in a wide range of critical research on cancer and vaccines. It also runs the supply and logistics network for the government’s remote McMurdo Station research outpost in Antarctica, moving people and supplies back and forth from the U.S. for a residential population that can exceed 1,000 in the summer. CEO Krone learned firsthand just how remote the base can be when he got stuck there over Thanksgiving in 2017.

Souvenirs of that longer-than-expected stay are the hundreds of photographs he took of the seals there, one of which adorns the wall opposite the desk in his office. Nearby, a somewhat cluttered bookcase contains mementos of aircraft he helped design and the Lockheed Martin deal, alongside childhood artwork from his three now-grown children.

When it was first put in the water in 2016, Sea Hunter was a slick gray beast, fierce-looking and intentionally tough to board. The ship lacked not just the interior amenities to house a crew, like sleeping quarters, a galley, and bathrooms, but also handrails along the sides and padding on the deck for traction. The Navy, after all, had asked for an autonomous ship that could track enemy submarines and resist boarders. But when the testers from Leidos launched its very first trips along the Columbia River in Oregon, it became apparent that they needed to add handrails and an anti-skid coating on the deck for safer human boarding. There’s also a small, bolted-on pilot’s cabin for shelter and some metal rails for connecting gear. Cook, the senior program manager, says some of the additions make him cringe. “It’s like a roof rack on a Corvette,” he says.

But without them, it would have been all but impossible for the engineers to come aboard and fix the engine two years later, while tossing on the high seas. In under an hour, Crabtree and the Navy engineers restarted the craft, tracing the problem to an easily corrected software setting.

While the airborne drones commonly used by the military are piloted by remote control, and some autonomous under­water craft use computer-controlled collision avoidance programs, Sea Hunter was designed to achieve an even higher level of self-control—a challenge not unlike that designing autonomous vehicles. Though sea traffic is nowhere near that of highway driving, the stakes of an error are significantly higher. And there are no road signs, traffic lanes, or dividing lines for the software to track. Cook, a self-described “autonomy snob,” says, “I think a [self-driving] car is easier.”

Leidos designed Sea Hunter to meet the fundamental rules of human ship-to-ship encounters, which require that a ship follow different procedures depending on its features and functions. Typically, one ship is to stay on course and the other is to give way. But the priorities differ for sailboats vs. powerboats, the direction of the wind, and many other criteria. Sea Hunter uses sensor data from cameras and radar to assess any other craft it encounters and properly choose the correct maneuver.

No hands on deck

The specs for the Navy’s Sea Hunter, and the expectations that surround it.

Length: 132 Feet Top Speed: 27 Knots The boat is designed to test a range of missions carried out by medium-size craft, such as tracking submarines with towed arrays of sensors or acting as a communications relay equipped with long antennas. Not as fast as some of the Navy’s newest manned ships, but speedy enough to keep up with a sub. For some missions, like clearing mines, speed is less important than the fact that the ship carries no crew. Range: 10,000 Nautical Miles Weight: 135 Tons Some of the largest savings achieved by unmanned vessels come from long missions. Sea Hunter could remain at sea for weeks, voyaging from California to Hawaii and back almost twice without returning to base. The fiberglass hulled boat isn’t meant for the front lines of battle but could serve as a prototype for future autonomous ships built with a variety of materials and missions in mind.

 

It was the Navy that sought the big test—an ocean crossing with “no human hands on”—to prove that the concept of unmanned vessels was ready for a much bigger push. After Sea Hunter passed with flying colors, the Navy Department issued requests in April for the design of truly combat-ready medium-size and large-size (up to 300 feet long) unmanned surface vessels. Says Rear Adm. Ronald Boxall, director of surface warfare for the Navy: “We’re looking for a mix of ships that gives us the most lethality per dollar.” Unmanned ships are “in a research and development phase right now, but they could cross into an operational procurement phase relatively quickly when we think we’re ready.”

For now, autonomous vessels are part of the Navy’s strategy to address the twin threats of the expanding Russian and Chinese navies. China is building a vast armada of surface and underwater craft as it tries to win dominance in the Pacific. The Russian military doesn’t have the same resources but is building a fleet of quieter and more efficient subs that could sneak around the world to deliver conventional or nuclear payloads. Sea Hunter, which carries no weapons, is designed to monitor these fleets, as well as to clear mines and provide a secure communications relay for the Navy’s largest warships.

In December 2017, the Navy ordered a second Sea Hunter from Leidos, being built in Gulfport, Miss. Next, the company will compete for a part in the 2020 medium and large unmanned vessel programs. It’s likely to partner with other contractors more expert in the world of shipbuilding, such as General Dynamics and Huntington Ingalls. That would be similar to Leidos’s work building imaging and sensor instruments for planes and guidance systems for cruise missiles, which are built by others.

A scientist at the Leidos-run National Cryo-Electron Microscopy Facility. | Photograph by Scott Suchman

Boeing and Lockheed Martin, for their part, have concentrated on underwater unmanned craft, avoiding the complications of navigating amid other vessels on the surface. And Rolls-Royce Holdings showed off renderings of an autonomous naval vessel somewhat like the Sea Hunter in 2017 but never produced a craft. It has since sold its commercial boating business to Norway’s Kongsberg Gruppen ASA. Kongsberg has so far focused on developing civilian unmanned craft. It has a refit ferry that navigated its way on a journey of a few miles around Finland’s Turku Archipelago and is also working with shipbuilder Vard Holdings to build a huge autonomous container ship that should be ready to sail next year.

Autonomous vessels will save a ton of money for the Navy. According to a study produced for the Defense Advanced Research Projects Agency, or DARPA, which initially oversaw the autonomous vessel program, Sea Hunter can operate for $20,000 per day, compared with $700,000 to run a fully manned destroyer performing similar missions. And with no sailors at risk, an autonomous fleet could serve as “pawns” for tracking subs, clearing mines, and acting as communications relays while manned vessels remain the “king” and “queen” pieces for large-scale battles in the Navy’s ocean-borne chess match against China and Russia.

For the unmanned Navy project, Leidos engineers ran simulations of more typical single- and twin-hull designs, as well as some submersible possibilities. But to their surprise, they found that a main hull with two outriggers was more stable, faster, and cheaper to maintain. “I think because we weren’t a shipbuilder, we really came at it with a very fresh look,” says Krone.

All three Leidos divisions are in healthy shape. Defense revenue grew a robust 7% in the first quarter of this year; the civilian unit, which makes up a third of revenue, was up 2%; and health care was the strongest segment, posting a 9% revenue gain. This unit, which contributes 18% of revenue, targets Medicare fraud and provides disability exams, but the crown jewel of the business is the management of the Frederick National Laboratory for Cancer Research in Maryland, which sports a $540 million annual budget.

The gleaming national lab sits on a serene hillside at the foot of the Catoctin Mountains, 45 miles north of Washington, D.C. Split off from nearby Fort Detrick, home of the nation’s bioweapons research, by President Nixon in 1972, the lab’s charge is to focus on cancer, AIDS, and other areas that have proved too tough or too uncertain to be profitable for the private sector. Leidos won a $5 billion contract in 2008 to run the lab and added a $1.5 billion extension in 2015.

In one darkened lab room in the ­bowels of the 330,000-square-foot facility, “Tommy’s dancing molecules” are getting zapped with laser light in a high-powered microscope. Appearing as zigzagging dots across a black display, the molecules are RAS proteins inside of cancer cells. Mutations of the gene that encodes the instructions for making the protein are at the root of 30% of all human cancers. “Tommy” is Tommy Turbyville. A scientist working for Leidos, he is trying to figure out if there’s a way to directly target the mutant proteins, which cause some of the deadliest forms of the disease, including cancer in the pancreas, colon, and lungs. The discovery of a drug that inhibits RAS could save millions of lives, but the private sector, which has come up empty after 30 years, has largely given up pursuing it on its own.

With a trim white beard and black glasses, and wearing jeans under his lab coat, Turbyville is full of energy as he bounds across the lab to explain that by tracking the dancing molecules, measuring their speed, and creating computer models for how they move, the project aims to uncover new vulnerabilities in mutant RAS that could be attacked with drugs. In another part of the lab, a $1 million robotic setup is injecting different compounds into test plates of RAS proteins.

A $7 million microscope used in cancer research at The Frederick National Laboratory, run by Leidos. | Photograph by Scott Suchman

Leidos scientists also operate a $7 million cryo-electron microscope that cancer researchers all over the country can use for free. Another project is focused on finding a way to lower the required dosage—and cost—of administering the HPV vaccine.

“It’s the perfect example of what a national lab should be doing,” says Len Freedman, chief scientist at Leidos’s biomedical research subsidiary. “RAS is behind some of the most common cancers, but despite thunderous efforts, nobody has gotten close to [designing] an inhibitor.” Still, the lab’s efforts are starting to bear fruit. Clinical trials for humans are starting this year for several promising drugs to address some RAS-related cancers, though it’s unknown whether the trials will succeed.

Back in San Diego, Sea Hunter spends most of its time these days in dock, going out to test new tweaks to its hardware and software about once a month. In person, the ship is larger than it looks in pictures—nearly half the length of a football field—and more fierce, with its two pointed outriggers and sharp bow. There’s a small plastic “good luck” hula girl in the cockpit but almost no other human touches. Visitors without clearance aren’t allowed to see what’s below deck, although it’s obviously not crew quarters. “That’s where the unobtainium time machine is,” CEO Krone jokes later.

The biggest threat to the ship these days is the occasional loafing sea lion that clambers onto one of the outriggers and won’t be moved. “You just have to wait until they get off,” says Cook, smiling in the California sunshine, while giant destroyers and cargo ships ply the blue waters of the bay and cruise past the famed Point Loma Lighthouse nearby.

This article appears in the June 2019 issue of Fortune with the headline “Adventure on the High Seas.”

More must-read stories from Fortune:

—The 2019 Fortune 500 list demonstrates the prize of size

—These businesses are making bets on a quantum leap

CVS wants to make your drugstore your doctor

—The Occidental-Anadarko merger reveals the crude truth about oil prices

—How AT&T has become a new kind of media company

Get Fortune’s Eye on A.I. newsletter, where artificial intelligence meets industry