Sky Cowboys: Cambridge’s Aurora Studies Ways to Lasso Robot Planes In Flight

11/25/08Follow @wroush

Here’s an idea worthy of Tom Swift: Send a big, long-range unmanned aerial vehicle (UAV) such as Northrup Grumman’s Global Hawk into a battle area with a flock of smaller, bird-sized robot planes in its belly. Jettison the baby planes for short-range, low-altitude, low-speed reconnaissance missions, then lower a cable to recapture them, one by one, before bringing the whole flock home.

That’s the scenario envisioned by engineers at MIT and Aurora Flight Sciences, a Manassas, VA, defense contractor with a major R&D lab in Cambridge, MA. Last week Aurora won a Small Business Innnovative Research contract from the Air Force Office of Scientific Research (AFOSR) for an initial study of the concept, which hinges on the development of an innovative cable system for retrieving so-called micro air vehicles (MAVs). The cable could be the solution to the biggest problem in aerial recovery scenarios: the speed mismatch between large and small craft.

“If you have a larger UAV, it can fly very fast for long distances, but it can’t fly slowly and look around in a city, for example,” says James Peverill, an embedded systems engineer at Aurora’s Cambridge lab. “But a smaller UAV can go down and look around at things more carefully. If you combine those two regimes, you can bring about a new capability.” The issue is that that “you can’t dock the two planes without some additional work,” says Peverill, because they can’t match speeds the way a fighter jet and a refueling tanker can.

You might not think that simply lowering a cable from a larger UAV would help, since the end of the cable would be moving just as fast as the mother ship. But Peverill and colleagues in the laboratory of MIT Department of Aeronautics and Astronautics are investigating a twist on the idea—literally.

If the larger craft flies in circles, Peverill and his colleague believe, the circular motion, together with aerodynamic drag, will cause the lower end of the cable to trace a smaller circle—meaning that “the end of the cable will be traveling slower than the large UAV,” he says. So much slower, in fact, that a MAV could approach and dock with it, in the same way a fighter hooks into the drogue basket of a tanker’s refueling hose.

The approach, code-named Sky Cowboy, has never been tried with robotic vehicles, according to Peverill—but it’s likely to be less violent and less potentially damaging than other ideas for air-to-air retrieval, such as having a large UAV fly up behind an MAV and snag it with a hook.

Aurora’s nine-month Phase I grant of about $100,000 will allow Aurora and its research partner for Sky Cowboy, MIT Aero/Astro professor Jonathan How, to test the idea on a small scale using the Real-time indoor Autonomous Vehicle test Environment (RAVEN), a motion-capture facility at MIT’s Aerospace Controls Laboratory. (We last wrote about How when he was helping a team of students build MIT’s DARPA Urban Challenge robot car.) You’ve probably seen “making-of” videos about digitally animated Hollywood movies like The Polar Express or Beowulf, where actors dress up in body suits covered with targets and their recorded movements are used to guide the motion of digital models. RAVEN does the same thing with model aircraft.

“It’s very expensive and difficult to instrument a radio-controlled plane to know where it is in a room,” says Peverill. “But if you use the motion-capture system, you can know exactly where it is without adding anything to the plane except the targets.” For the Sky Cowboy tests, targets will also be attached to a cable dangling from a radio-controlled plane that’s flying in circles, allowing researchers to measure whether the end of the cable behaves as predicted.

The RAVEN data may also be used to construct a digital model that could enable the team to explore various configurations for the cable, says Peverill. If the results are encouraging, they could help Aurora lasso a much larger Phase 2 grant to fund development of a full-scale prototype system.

Wade Roush is a contributing editor at Xconomy. Follow @wroush

By posting a comment, you agree to our terms and conditions.