Quantcast
Building Airplanes That Fly Themselves, On A Red Bull and Cigarettes Bender

When it comes to hacking, MIT grads Andrew Meyer and Okie Jon Williams are the genuine article.

When it comes to hacking, MIT grads Andrew Meyer and Okie Jon Williams are the genuine article. On a diet of cigarettes, Red Bull, and McDonald’s (no sign of Four Loko) these mad circuit-obsessed mavericks managed to put together a robotic plane that can fly itself.

To get away from the stiff movements of C-3PO-style robots, they implemented a machine learning system for movement using the same algorithms used for computer vision, speech to text, and, oddly enough, spam filtering. It’s somehow appropriate that the tools that help keep Viagra ads and Nigerian get-rich-quick schemes out of your email can help fly a plane.

It’s the kind of thing that the military might want to get in on – or perhaps those terrorist hackers.

While the duo aims to produce a plane capable of “autonomous acrobatics”, they are also making a standardized interface for amateur roboticists to start making prototypes without much hardware experience. As Nadja Oertelt wrote on VBS,

They’ve basically taken a socialist approach to powerful autonomous computing, with the ultimate aim being to bring the advances of the robotics industry within reach of the little guy, though in this case that’s less a reference to the lumpen prole than the kids currently experimenting with batteries, wire, stolen matches, and circuitry while everyone else is playing kickball—the young nerds of tomorrow.

We hung out with them last year, and fortunately for you, brought our video cameras.

Check out their company, Leaf Labs, and see our films on other robot hackers, like NYC Resistor, Yoshiyuki Sankai, and a pair of Sumo robot builders.