AR Drone Hackathon
The Parrot AR Drone is the venerable little $300 toy you can buy from a Brookstone store, unobox, charge, and chase your cat around the house. But because it provides easy entry into the drone hobby, complete with high-definition camera and a decent suite of sensors and accelerometers to make navigation a breeze, it’s also been catnip for savvy hackers and developers.
Tim Pool first tried his hand at hacking the AR Drone in January 2012, attempting to develop a revolutionary platform for the world observe the Occupy Wall Street protests. Pool and his associates had some success, but were hampered an inadequate Software Development Kit (SDK), and the inherent shortcomings of the lightweight, low-powered $300 toy.
Circumstances have changed since then. The Parrot company has released its second version of the AR Drone, which now features improved air pressure sensors, a magnetometer, and improved camera resolution. There’s also an improved SDK and a dedicated community of people who try to push the toy to its limit.
On October 5, a group of 60 developers got together at the historic Stadtbad Oderberger bathouse in Berlin (an architecturally stunning building that’s no longer used for public bathing) for an all-day hackathon called Nodecopter.js to see exactly what could be done with the AR Drone.
“Unlike the firmware, the client protocol is open, and Parrot publishes an SDK
(signup required to download) including a good amount of documentation
and C code. Their target audience seems to be mobile developers who can
use this SDK to create games and other apps for people to have more fun
with their drones,” wrote CrashingDutchman on the DIY Drones website. “However, the protocol can also be used to receive video and sensor
data, enabling developers to write autonomous programs for the upcoming
robot revolution.”
What were they able to accomplish on that day? Some hackers made up innovative new controls for the ‘copter, including one system that could operate the drone via one thumb and a Nintendo Wii nunchuck controller. Some got the drone to obey voice commands.
Several teams attempted image recognition using the drone’s main camera, with some successfully pulling off facial recognition.
And then another made the drone charge after anything with a red hue, much like a matador’s bull.
Simon Madine has a great write-up and photos on Creative.js.
