A doctoral student has developed a self-contained, onboard processing system to allow a drone to fly around and beneath obstacles while traveling at more than 30mph. Andrew Barry’s “stereo-vision” algorithm runs off two cell-phone-sized processors, allowing a flying drone to map its path on the go at a rate 20 times faster than current software. This video shows the drone in action, followed by a chase drone.
The drone used in the test weighs just over a pound with a 34-inch wingspan, and was made from off-the-shelf components costing about $1,700, including a camera on each wing, MIT said in a press release. Credit: YouTube/MIT Computer Science and Artificial Intelligence Laboratory