IEEE Future Directions
Related Articles
I reported some time ago of an artificial eye, made by curved sensors, that was mimicking an insect eye. Now I run onto a news where this artificial eye has been used to provide information on the environment to a drone-robot to avoid obstacle.
Biorobotic researchers at the Institute des Science du Mouvement, University of Marseille, France, have used the sensors aggregated onto two semi-circular bended layers, mimicking dragonfly eyes to provide the robot with information on the terrain making it able to keep a safe distance from it. You can take a look at the video clip.
The robot, named Beerobot, is using a vision algorithm similar (in its effects) to the one used by flying insect. The information captured by the eye are processed as an "optic flow" where what lies in front is basically fixed whilst what is on the edges "flies by" with a progressively increased speed depending on the angle (with the maximum speed reached by object located at 90° from the front). The processing of the optical flow provides the same information that would be provided by an accelerometer, a speed indicator and an altitude indicator (since the optic flow is in the 3 dimension, a sort of "cone" that the insect uses as a beacon.
It is an amazingly simple process in terms of computation effectiveness, and it works pretty well in keeping the insect (and the robot drone) in the desired flight path avoiding obstacles. If there is an increase in flow velocity it means that the speed of the insect has increased or an object is getting closer. Since the speed in flight is basically constant it is most likely that an object is getting closer. The flow velocity is easily assessed by the time it takes to a detail in the optic flow to move from one sensor to the next one.
The BeeRobot brain consists of three feedback loop: altitude, speed and stabilisation. These feedback loops are all depending on the optic flow and are all it takes to fly and avoid obstacles.
Notice how Nature has brilliantly solved the flight navigation without the use of accelerometers and complex processing of data. Quite the contrary of our airplane that relies on accelerometer and speed measure (pitots) with data crunched by powerful flight management systems to fly.
