Sunday, November 26, 2006

MIT 'Air Force' could help perfect Unmanned Craft by Peter J. Howe

Who says battery-powered airplanes have to be outdoor toys?

Not aeronautics professor Jonathan How of the Massachusetts Institute of Technology , who along with a team of students this fall has turned an MIT lab into a first-of-its-kind US test bed for "unmanned aerial vehicles" that, with the help of computers, fly themselves.

It's undeniably fun, How admits, to get away with flying a model helicopter inside. But his team's work, sponsored by aircraft giant Boeing Co.'s Phantom Works research unit, could one day help revolutionize one of the fastest-growing sectors of the aviation industry, remote-controlled flying devices that are increasingly being used for everything from warfare and border surveillance to battling forest fires and doing seismic testing for oil deposits.

Teal Group , an aerospace and defense market-analysis firm in Fairfax, Va., recently projected that worldwide spending on unmanned aerial vehicles and related systems will represent a $55 billion worldwide market over the next 10 years. Annual spending on flying drone systems could triple, to $8.3 billion in 2016 from $2.7 billion now.

The new MIT indoor flying lab is helping to simplify one of the biggest challenges to wider deployment of unmanned vehicles: developing the very complex, perfectly reliable software and telecommunications systems to manage a fleet of flying devices and keep them from crashing into each other.

"Ultimately, when you are taking these devices out into real-world applications, you want people to perform a task like surveillance of the border. You don't want them spending a lot of time figuring out how to fly the vehicle," How said.

To test and debug a multiple-vehicle flying system outdoors normally requires four people monitoring every vehicle, How said, or potentially over three dozen people to run a test of 10 flying drones. "That is logistically hard, and very costly," he said.

With the MIT system, not only can one person handle several flying devices at once, "You can have a student essentially operate this from their bedroom," through a high-speed Internet connection.

How's air force consists of a half-dozen four-rotor helicopters, each about the size of a chicken and costing around $700. Their actual moment-to-moment flying is controlled by a network of computers.

So far, researchers have been able to complete tasks like landing a mini-chopper on a motorized toy truck, a good simulation of landing a drone on a Humvee in the desert or a battleship at sea.

Their next milestone is to keep a fleet flying for seven straight days, which requires helicopters flying back to a landing pad to recharge their batteries. Several graduate students in electrical engineering and aeronautics, including Brett Bethke, Daniel Dale, and Mario Valenti, handle much of the nuts-and-bolts work of keeping the fleet flying.

To create the equivalent of an indoor satellite positioning system, How's lab uses the same motion-monitoring systems from Vicon , a British technology company, that Hollywood studios have used for animated films. For a cartoon movie, the systems track an actual moving human to generate realistic-looking animated character movement. At MIT, they perform toy-helicopter air traffic control by tracking their position to within one-tenth of a millimeter in any direction.

The work directly addresses some of the major obstacles to wider use of unmanned aerial vehicles, said John Vian , a technical fellow with Boeing Phantom Works. "Enabling complex and coupled systems to operate reliably is really the biggest challenge we're facing," he said. "We need smart systems, and Jon How and the folks at MIT have the capability to make them work."

Boeing currently has "a stretch goal," Vian said, of coming up with a system that can enable one operator to control 100 vehicles. That will mean solving all kinds of nitty-gritty problems involving computer software, mathematics, motion monitoring, and communications, Vian said.

But for How and his fellow researchers, Vian added, it won't be entirely boring.

"You can see from what's going on in the lab," Vian said, "that it's just a blast."

No comments: