University of Zurich researchers built an AI that flies around on a drone telling it how to navigate the streets thanks to data gleaned from self-driving cars and GoPro-toting bicycles. (Anyone else picturing Atreyu flying around on Falkor the luck dragon? Just me?)
In a new paper appearing in IEEE Robotics and Automation Letters, Davide Scaramuzza and PhD student Antonio Loquercio laid out the work they’ve done training an autonomous drone to fly through the streets of a city using data collected by cars and bicycles. The AI-equipped drone takes that data, builds a map, locates itself on the map, and then plans a route, using images from a camera to provide speed and steering commands, follow road markings, and avoid collisions. No luck (or luck dragon) necessary. It’s a pretty cool way for machines to learn from each other, which will be necessary when humans are eventually enslaved by our machine masters.