Autonomous Drone Sees The World Through Smartphone’s Eyes

This tiny unmanned copter will take your phone to heights it’s never seen.

One continuing challenge in building autonomous drones has been getting them to figure out where they are in space. The quantitative measures they should see the world determines the precision with which researchers can control them. And so far the efforts of researchers has led to a bunch of different systems that, though sometimes controllable by inexpensive, commonplace devices like smartphones, have a limited ability to inform each other in their common mission because of their expensive, closed systems. Now comes the SmartCopter, developed by graduate student Annette Mossel and her fellow researchers at the Vienna University of Technology. It’s a drone that puts a ubiquitous smartphone onboard, allowing the phone to use its processing power in perceiving the environment.

One of the researchers’ other goals was to create a device that can navigate over very precise distances, particularly indoors. This means that the smartphone’s orienting technologies, such as its magnetometer, accelerometer and GPS must be supplemented; according to the U.S. Government website about GPS, it’s accurate only within a distance of 26 feet. To solve this problem, the smartphone is supplemented with two off-the-shelf web cameras on the drone, which, with the help of an open-source, vision-based platform called PIXHAWK, work together stereoscopically to process visual information about the environment, much like human eyes. Among possible users for this drone, researchers count surveying disaster sites, observing the condition of walls in large, open spaces like churches, and helping people navigate indoor areas such as shopping malls.

The drone still has many limitations that will keep it inside the lab for the time being. Its sense of orientation across a horizontal distance is determined in a fairly crude way, using paper markers that the drone observes from a pre-determined height. Future incarnations might use the smartphone to track features of a room like corners and gradients so that the drone can map out a room without prior intervention. And though a smartphone-based device like the SmartCopter could potentially be cheaper to one day purchase because you supply the CPU, this experimental model, cobbled together from off-the-shelf items like webcams, didn’t yet prove that. Still, this promising device (from the team of a rare female engineering student, no less) could one day save lives.

Annette Mossel

Image: FastCo Labs

Sources: FastCo Labs, original research paper, MIT Technology Review

Quantcast