Accelerometers have become integral components for many of our favorite gadgets. By measuring acceleration forces, such as gravity or someone's arm waving clumsily back and forth, these sensors can accurately identify a device's angle in relation to the Earth. It's how your smartphone knows when to automatically switch between portrait and landscape orientation. Now, scientists are researching how drones can be built to fly autonomously without the use of accelerometers. It's led to the creation of "BeeRotor," which, as the name implies, takes inspiration from the visual cues and analysis used by winged insects.
The approach is called "optic flow," and it measures both distance and elevation based on how your eyes naturally interpret movement. So when you're cycling at high speed, the landscape on the horizon looks relatively stable; if you move your head to either side, however, the scenery rushes by faster and faster, topping out when your noodle is turned at exactly 90-degrees. The BeeRotor recreates this effect with 24 photodiodes that record contrasts and their motion in the environment. When a section of the terrain moves from one sensor to another, the robot uses this data to calculate the angle at which the scenery is passing by, and by extension, its own position. Likewise, the BeeRotor can keep tabs on its speed by analysing how quickly the landscape is moving across its "eye."
Researchers from the Institut des Sciences du Mouvement Etienne-Jules Marey are using optic flow to develop three stabilisation-focused feedback loops for BeeRotor. The first automatically adjusts the robot's altitude in accordance with the floor or roof. The second changes BeeRotor's speed depending on the size of the space it's flying in. The final loop moves the robot into position so that its "eye" always has the best possible view of the approaching terrain. In the video below, a tethered BeeRotor is able to safely traverse artificial tunnels that change in size and elevation. Scientists hope to develop the technology further into a lightweight replacement for accelerometers on smaller drones, as well as a backup system for larger models carrying out important research.
[Image Credit: © Expert & Ruffier (ISM, CNRS/AMU)]