A Couple Nerds Are Trying to Save the Navy’s Killer Drones

Rarely has such obscure research had such huge strategic implications

A Couple Nerds Are Trying to Save the Navy’s Killer Drones A Couple Nerds Are Trying to Save the Navy’s Killer Drones

Uncategorized September 6, 2013

An X-47B drone on the USS George H.W. Bush in May. Navy photo A Couple Nerds Are Trying to Save the Navy’s Killer Drones... A Couple Nerds Are Trying to Save the Navy’s Killer Drones
An X-47B drone on the USS George H.W. Bush in May. Navy photo

A Couple Nerds Are Trying to Save the Navy’s Killer Drones

Rarely has such obscure research had such huge strategic implications

The U.S. Navy had a vision. A robotic vision, in which swarms of super-smart, far-flying drone warplanes ranged across the vast Pacific Ocean from steel aircraft-carrier decks, extending the reach of American military might against the growing arsenal of an increasingly belligerent China.

Fighting war in the air at great distance and without needlessly exposing pilots to danger. Helping protect the manned planes that would still comprise the bulk of the Navy’s forces.

That vision came close to becoming a reality, when this summer the experimental X-47B drone blasted off the deck of the carrier USS George H.W. Bush off the Maryland coast—the first at-sea launch of a representative robot warplane.

But no sooner had the X-47B completed its first round of ocean trials than the Navy began pulling back from its robotic vision, seeming to lost its institutional nerve as it announced a profound downgrade of the carrier drone’s capabilities.

Where once the ‘bot had been meant for full-scale aerial warfare in the face of stiff enemy defenses, now it would be lighter, slower, less stealthy—optimized for routine patrols in undefended airspace instead of intensive aerial combat.

And rather than joining the Navy’s manned F/A-18 fighters and E-2 radar planes on the decks and in the maintenance hangars of the sailing branch’s 10 nuclear-powered flattops, the drones would be kept separate from the daily tussle of deck ops. The ‘bots would launch from the decks only at night, after the other planes had been tied down and their pilots tucked into bed—in other words, when the carriers weren’t so busy.

The change of flying schedule meant the drones would never be included in the Navy’s main air-warfare plans. They would be second-stringers, deliberately held back from the most difficult fighting. The drones were being denied the chance to prove themselves.

And for one major reason: the Navy was scared. Scared of how drones would blend with manned planes on busy carrier decks. Scared that an autonomous aircraft might lurch out of control, damage the ship and other planes—even kill crewmen. Safe “deck-handling,” as the practice is called, was one of the drone’s biggest problems, as far as the Navy was concerned.

The brass sequestered the ‘bots to the nighttime hours partly in order to give them a relatively empty deck on which to maneuver. The change might have reduced the chance of a robotic accident, but the increased safety came at the cost of the Navy’s most promising new technology, a new weapon very nearly ready for combat that could shift the balance of power away from a rising China and back to the Navy’s advantage.

It’s not too late to fully deploy the drones. In the fluorescent-lit halls of one of the country’s leading robotics lab, just outside Boston, researchers are scrambling to devise protocols—rules, really—for mixing people and robots in the same busy space, just like a carrier deck.

These safety protocols, under development by Missy Cummings and Jason Ryan at the Massachusetts Institute of Technology, could be exactly what the Navy needs to reverse its earlier decision … and unleash its future drone warplanes to fight alongside manned fighters in any future air campaign.

That’s right: these nerds could be the key to mobilizing the decisive robotic warriors of the next high-tech war.

An X-47B comes in for a touch-and-go landing on the USS George H.W. Bush in May. Navy photo

Dangerous ballet

Seventy planes including 40 jet fighters. Scores of sailors on foot or steering tugs, fuel trucks and trailers stacked with live bombs and missiles. A hundred or more takeoffs and landings in a typical 12-hour daytime shift. The 4.5-acre steel deck of a Navy aircraft carrier is one of the busiest places in the world.

The Navy’s own literature described the carrier’s hectic topside as a “ballet.” A potentially deadly one. “The dance floor is a hot, stench-filled, steel deck that can be measured in acres and contains hundreds of hazards,” the sailing branch’s safety magazine warned.

Planes hit other planes … and sometime people. Sailors get blasted off the deck and into the water. Machinery malfunctions. Fires start. In 2008 an F/A-18 struck and killed a sailor on takeoff on the USS Dwight D. Eisenhower. Thirty-four other sailors have died on flattop flight decks since 1980—and even more before that. A 1969 fire killed 24 crew on the USS Enterprise. Explosions on the USS Bennington in 1954 resulted in 91 deaths.

The topside dangers were very much in the minds of Navy brass as they began planning the introduction of the fleet’s first carrier-launched Unmanned Aerial Vehicles. The roll-out began with two bat-shaped test drones built by Northrop Grumman. The pilotless X-47Bs—62 feet from wingtip to wingtip, 22 tons fully loaded—follow GPS coordinates in the air; while taxiing on the ground, they are steered by someone carrying a bulky, awkward, arm-mounted remote control.

The X-47Bs flew from land for a year before heading to sea for non-flying tests aboard the carrier USS Truman, sailing off the East Coast. The Truman had practically no other planes on board at the time, giving the tests the widest possible safety margin. “Every evolution with this aircraft is taken step-by-step because we don’t fully know how it will react to a carrier environment,” said Lt. Cdr. Larry Tarver.

In May the X-47Bs staged aboard the carrier Bush for steam-catapult takeoffs, followed by arrestor-wire landings on the Bush in July. Again, there were virtually no other aircraft on the flattop at the time.

The summer trials were successful, but by then the Navy had made up its mind: despite drones’ promise of lifesaving autonomy in combat, the sailing branch did not trust the ‘bots to fully blend with manned planes on a crowded deck. Especially as pressure mounts to make robots more autonomous in their decision-making.

Planning documents, leaked to the press in June, revealed that the Navy would permanently separate drones from the rest of the carrier air wing once combat-grade robots—successors of the X-47s—started entering service around 2020.

Manned planes like the Hornet fighters would fly during the day, the final planes landing as the sun went down. The last airplane launched before the end of the normal deck cycle would be a drone. It would fly for 12 hours, scouting ahead of the carrier and even dropping a couple small bombs if it found a suitable target, before returning to the flattop at dawn before regular manned flying resumed.

To that end, the combat drone would be lighter, less capacious and less able to avoid enemy detection than originally planned. It would lack the fighting qualities of the latest manned warplanes because it would not be meant for major combat.

Officers took pains to portray the manned-unmanned division as a good thing, pointing out that the drones nighttime-only schedule could “span that gap while your flight deck is shut down,” in the words of Capt. Chris Corgnati, a Navy drone planner.

But Bob Work, who recently stepped down as Navy undersecretary, called the separation for what it really was—a sign of Navy “reticence” in the face of unmanned planes’ vast potential. “You completely decouple truly integrated operations,” Work told the U.S. Naval Institute’s blog. “In my view, it’s entirely the wrong way to look at this.”

Instead, drones should be treated as equals of manned planes and be included in the main, daytime mission planning—or so Work’s thinking goes. But that means reassuring the Navy that flying robots and traditional, manned warplanes really can work together safely on carrier decks.

And that’s where MIT nerds Cummings and Ryan come in. They’re turning drone safety into a literal science—one that could help restore the Navy’s killer drones to their rightful place among the other carrier planes.

Ryan and Cummings’ simulation. Note the diamond-shaped drones on deck, and the colored circles representing people. Jason Ryan image

Researchers to the rescue

Cummings—gregarious, blonde, a pioneering woman F/A-18 pilot in the 1980s and a one-time guest of both Jon Stewart and Stephen Colbert on their respective Comedy Central talk shows—heads the Humans and Automation Lab at MIT and also manages research projects for the Navy.

The dark-haired Ryan, talkative and energetic, who according to his bio enjoys sports, music, history and cooking, is one of Cummings’ PhD students. Together Cummings and Ryan explore the ways people and robots interact.

Think programming of automated warehouse forklifts. Or collision-detection warnings for drones flying in commercial airspace. With more and more robots crawling, swimming and flying among us, this interaction is becoming a serious concern. “We want to release these things [robots] into human-centric systems full of squishy bags of meat [people],” Ryan says. “What happens when there are 30 robots and 50 people in the same space?”

Is the biggest hazard the lag-time between an automaton detecting a person in its path and moving to avoid? Is evasive action itself a danger, as robots veer wildly to avoid hitting people—and then collide with each other or people who weren’t originally in the way?

How should robots and human beings interact? What safety rules should be enforced? “Should there be temporal or spatial divisions between manned and unmanned?” Cummings asks rhetorically. That is, should ‘bots move only during set-aside periods of time, during which people stand back? Or should the machines have their own designated lanes, into which human don’t cross?

Must carrier drones always be steered on the ground by a sailor with an unwieldy remote control, or can ‘bots be trusted with more autonomy, free to steer themselves based on input from sensors?

No one knows for sure. Hence the Navy’s reluctance to fully integrate killer drones on carrier decks during the busiest phase of the flying cycle. Testing is the only way to begin to understand human-robot interactions—and the only way to get the sailing branch to change its mind.

But testing in the real world is expensive. In the Navy’s case, a carrier drone trial means setting aside a $7-billion flattop and all its airplanes, each costing $50 million apiece or more—plus redirecting the labor of thousands of sailors each earning hundreds of dollars per day. Building and testing the two X-47Bs since 2007 has cost a billion dollars.

A billion dollars for just a few days at sea—and a relative paucity of data.

A better understanding of man-drone interactions requires more testing. Lots more. But that can only happen if the testing is cheaper. So with Navy funding Cummings and Ryan are working to create a smart computer simulation that can cram a bunch of code-based drones, manned planes and sailors on a mock carrier deck, set them in motion … and see what happens.

Over and over, as fast as a bunch of Navy developers can boot up their laptops and click a few buttons, planes made of pixels will maneuver on graphical carrier decks, dodging tiny colored circles representing sailors, backing onto catapults and launching into the air.

With its high-res graphics, user inputs and intricately-modeled artificial intelligence, the sim is “basically like every video game you’ve ever played,” Ryan tells War is Boring.

Ryan and Cummings’ program has taken a year to code; it’s nearly done. Soon the Navy will be able to put simulated drones with simulated manned planes and simulated people on simulated carriers—and hopefully figure out how to blend real piloted and real pilotless planes with Ryan’s so-called “squishy bags of meat.” That is, us.

The end result—hopefully, eventually—will be sea-based aerial robots restored to their full combat potential. Killer autonomous flying machines that zoom off and on the busy Navy flattops alongside the manned F/A-18s, ranging hundreds of miles in broad daylight to face the enemy head-on.

Drones that are free to fight with their full might.

If you have any problems viewing this article, please report it here.