9 years, 5 months ago

Andrew Barry's self-flying drone avoids obstacles at 30MPH

Drones are predicted to do everything from flying in swarms to pick out targets in war zones to delivering our post. Scroll down for video AUS engineer has developed an obstacle-detection system that allows a drone to autonomously dip, dart and dive through a tree-filled field at speeds exceeding 30mph 'Everyone is building drones these days, but nobody knows how to get them to stop running into things,' said Andrew Barry, a PhD student at MIT's Computer Science and Artificial Intelligence Lab who developed the system as part of his thesis. Barry's stereo-vision algorithm allows his drone to detect objects and build a full map of its surroundings in real-time, 20 times as fast as rival software. While traditional algorithms have used depth-field measurements provided by cameras to determine a drone's path, Barry said the approach is labour intensive and prevents it flying at more than 5mph without specialist software. The design includes a camera on each wing so the machine can 'see' in stereo and two computer processors to process the information, which are no better than those found in mobile phones Operating at 120 frames per second, Barry's software, which is open-source and available online, is able to extract depth information at a speed of 8.3 milliseconds per frame.

Daily Mail

Discover Related