Here's another great story from our friends at Popular Mechanics that looks at cutting edge research into drones that fly autonomously inside structures. That's something that until now could only be done (barely) by wheeled mini-bots. But as you can see from this report, engineers still have a long way to go.
It's not the most attractive spy bot, but the unmanned aerial vehicle hovering some 20 ft. away is doing its job. For now, that means staying right where it is, weaving ever so slightly under the weight of the webcam strapped to its back. There's nothing particularly interesting to look at with this UAV, a commercial four-rotor model that any RC hobbyist could put together. But no one is piloting this modified drone -- it's flying autonomously, stabilized a few feet above the floor of MIT's RAVEN lab. Like most of the aircraft tested here, this model is a puppet, receiving input not from onboard processors, but from a nearby computer.
As it continues to buzz in place, an array of 18 motion-capture cameras tracks the UAV, providing 3D positioning data to determine just how stable it is. Specifically, those baleful red cameras -- the same kind Hollywood visual effects teams use to transpose an actor's movements to a computer-generated counterpart -- are tracking the tiny Styrofoam balls attached to the drone. On the computer monitor, these balls show up in real time, mapping the UAV as a cluster of dots, swaying in midair. I'm somewhere between impressed and bored when the drone begins to drift. A second later and it slams into a plexiglass divider, as hard as a hockey player.
It will take some time to figure out why this little craft suddenly lost control. But that's the point of RAVEN, or Real-Time Indoor Autonomous Vehicle Test Environment, where geeks capture every flight -- and collision -- in painstaking detail. There are no accidents here, just problems that haven't been sufficiently analyzed. "RAVEN gives us the freedom to test whatever we can build," says Jonathan How, director of MIT's Aerospace Controls Lab. "And we can build wonderful things, even in 24 hours."
One of the researchers has done just that, and is now preparing to fly a drone that was redesigned, then cobbled together out of lightweight foam core. Of course, this isn't exactly the next generation of missile-packing Predators; the toylike creation in front of me, with its circular wing and miniature nose-mounted propeller, is more of a testbed than a prototype. All of the UAVs covering nearly every surface of this lab, from high-end RC planes the size of a small child to a store-bought flying insect produced by WowWee, are just tools to develop flight control algorithms for indoor robots.
As challenging as it is to make something fly itself, designing a drone that can function indoors is even harder. For an indoor UAV to meet all of the military's expectations, it would need to be able to fly into a building and find a suitable spot to perch and observe, all without relying on GPS contact. "The ultimate vehicle is a bat that you can download data from," How says. Bats have the ability to perch, plus echo location to detect obstacles, and the agility to keep from slamming into them.
At the moment, nothing in development can effectively pull off even one of these functions, much less all three. Researchers at Carnegie Mellon University, for example, are sticking to basic navigation, with a small robot helicopter that uses sonar and cameras to avoid bumping into obstacles while flying indoors. But the lab here has a novel approach: Instead of focusing on building better sensors or more powerful vehicle-mounted processors, researchers at RAVEN are fine-tuning the mechanics of autonomous aerobatics. The 18 motion-capture cameras provide a perfect sensing environment, and the dedicated computers in the lab, which communicate with the test drones via radio transmitters, provide the brainpower. "People might say our UAVs aren't autonomous," How says, "but in this environment, the entire system is autonomous."...