A Twin Dilemma

(December, 2004)

Entering the Machine Intelligence Lab in Benton Hall, I found myself face to face with the ‘Jasons,’ one seated, one standing, like some kind of stoic-faced twin dilemma there amid their stash of electronic components and workbenches. Not that they really look anything alike. Perhaps it’s just that after five years of working together they’ve taken on a similar air, because Jason Grzywma and Jason Plew build the stuff that goes inside the MAVs.

Grzywma shows me their latest designs, a small box containing an altimeter, a gyro, an accelerometer…

What started out as a pet project of Dr. Peter Ifju began to take on more momentum in the late nineties when a grad student developed a horizon-tracking algorithm. Then Dr. Michael Nechyba came on, a graduate of Carnegie Mellon’s famed Robotics Institute. Now his deal was computer vision, so it all kind of came together. The team began to incorporate computer vision and autonomous flight into the machines, to make it so that the MAVs could identify something like a car or a building on their own, to navigate. Or to track.

Enabling the MAVs to fly independently in an urban setting is still a long way off, but they’ve been able to achieve some autonomous flight in open areas with the larger vehicles.

On the MIL’s MAV website, you can see some footage of Ifju getting buzzed by his own creation—going down like the dog in Duck Hunt—from two perspectives: the ground, and the MAV’s.

Originally the MAV project got off the ground with NASA grants to Ifju and Nechyba’s labs. Currently, they receive funding from NASA, the Air Force Office of Scientific Research (their biggest endowment), and also DARPA and Eggland AFB.

I asked to hold the decrepit looking little six-inch, the black, sleek lines suggesting a mini-UFO, or integrals. Though Grzywma assured me that it was robust, having been in twenty crashes, I felt struck by the seeming fragility of the tattered latex wings...