Researchers Keep Damaged Drones Flying With Onboard Cameras

SEE ALSO: ENGINEERS TURN PINEAPPLE LEAVES INTO BIODEGRADABLE DRONE PARTS

"When one rotor fails, the drone begins to spin on itself like a ballerina," said Davide Scaramuzza, head of the Robotics and Perception Group at UZH. 

So once it starts spinning, the drone can no longer properly estimate its position, eventually leading it to crash. Sometimes GPS signals can help a drone stay stable, but when these are weak or unavailable, the drone crashes. 

The team instead turned to onboard cameras to provide visual information to the drone when a rotor failed. 

For its research, the team attached standard cameras and event cameras to a quadcopter. Standard cameras provided recordings of images taken several times per second at a fixed rate, while the event cameras used independent pixels activated when a change of lighting happens. 

Then the team developed an algorithm to combine the two cameras' information, using it to track the drone's position relative to its surroundings. This then allowed the onboard computer to control the drone as it flew, or indeed, spun, with only three rotors. In normal light conditions, the team found the drone functioned well, however in lower light conditions there were a few issues.

"When illumination decreases, however, standard cameras begin to experience motion blur that ultimately disorients the drone and crashes it, whereas event cameras also work well in very low light," said the first author of the study Sihao Sun, a postdoc in Scaramuzza’s lab.

Given how steadily drones are becoming a regular part of our lives, it's important that safety takes a top priority, not only for people potentially in the path of a falling drone, but also for all the collected data onboard the drones and the cost it takes to develop them.