Slashdot Mirror
MIT Researchers Develop Autonomous Indoor Robocopter
An anonymous reader writes "Researchers at MIT's Robust Robotics Group have developed a robotic helicopter capable of autonomously flying inside buildings or other GPS-denied environments. It has an on-board camera and a laser scanner that maps the local environment. The video talks about search-and-rescue and civil engineering applications, but it also brings somewhat scary reminders of Minority Report to my head. How long till I see one of these chasing me down a dark alley? The team's website has more videos showing earlier stages of the project."
Good luck with that.
This is interesting, although by using a quad-rotor helicopter, they seem to have mostly solved computer vision problem rather than a control system problem
Quad rotor and coaxial helicopters are very stable and have gotten pretty popular as entry-level helicopters because they are so easy to fly. The downside is that they don't really have the efficiency characteristics to fly outdoors like collective-pitch (e.g. like real full-size) helicopters.
Since the focus of the challenge was to fly indoors, using a quad-rotor is the natural choice. I'd like to have understood better how other teams failed and what kind of helicopters they used.
Collective-pitch helicopters, on the other hand, are extremely difficult to fly. Until the Stanford Autonomous Helicopter (http://heli.stanford.edu/), I believe no autonomous control system has been able to successfully fly one, even for very simple maneuvers.
Why a helicopter when you could use a blimp type device? Million times easier and more stable. And less fragile or dangerous.