If one of the motors conks out on a quadcopter drone, the aircraft usually just crashes. Thanks to new research, however, such disabled drones could soon not only remain airborne, but even complete their trip.
Typically, when one of a quadcopter’s four motors ceases functioning, the drone starts rapidly spinning, losing its orientation in three-dimensional space. And while there are systems that use GPS data to help reestablish that orientation, GPS signals aren’t always available – particularly if the drone is flying indoors.
Seeking an alternative, scientists from the University of Zurich and the Netherlands’ Delft University of Technology equipped a small quadcopter with two tiny onboard cameras. One of these was a regular optical camera and the other was an event camera – the latter type incorporates pixels that independently respond to changes in brightness as they occur.
When the copter began spinning after one of its propellers was removed, special algorithms were used to perform a real-time analysis of the output from the two cameras. By tracking how the location of specific visual reference points changed from one frame of video to the next, those algorithms were able to calculate how the aircraft’s location was changing. The drone was thus able to compensate for those changes, allowing it to maintain its position while still spinning – it could even complete a preprogrammed flight path.
When the optical camera was used on its own, its video became too blurry under low-light conditions, causing the quadcopter to crash. By adding in the more sensitive event camera, however, there was sufficient visual data to keep the drone flying even under dim lighting.
A paper on the research was recently published in the journal IEEE Robotics and Automation Letters.
You can see the technology being put to the test, in the video below. A previously developed system, created at the ETH Zurich research institute, allows quadcopters to safely land even when up to three of their motors are shut off.
Source: University of Zurich
Autonomous Quadrotor Flight despite Rotor Failure with Onboard Vision Sensors: Frames vs Events