Robotic Eye Inspired by Winged Insects Stabilizes a Robot's Flight
Reports of a recent study conducted at the Institut des Sciences du Mouvement --- Etienne Jules Marey (CNRS/Aix Marseille Université) indicate that biorobotics researchers have developed an aerial robot that is the first ever to have the ability of flying over an uneven topography using visual stabilization rather than an accelerometer. The results of the research study were published in the scientific journal Bioinspiration & Biomimetics on February 26th 2015.
The robot, known as BeeRotor, uses optic flow sensors the design of which was inspired by insect vision to avoid obstacles and adjust its flying speed. BeeRotor was tested by flying it along a narrow tunnel which had both uneven and moving walls. Its performance was excellent, moving quite easily without taking any measurements of either speed or altitude like other aircrafts do.
All kinds of manmade aircraft use an inbuilt inertial measurement unit (that includes accelerometers) to stabilize both roll and pitch in line with either the horizon or the direction of the Earth’s center. The accelerometer is used to take measurements of all the aircraft’s accelerations, including gravity. It is however important to note that there is no corresponding equivalent of this vital tool in insects, meaning that they fly quite naturally without having any need for recording this information.
Winged insects were therefore the major inspiring factor for researchers Franck Ruffier and Fabien Expert, the creators of the tethered flying robot. BeeRotor has a length of 47 centimeters and a weight of 80 grams and was able to autonomously avoid all vertical obstacles in the moving-wall tunnel. Ruffier and Expert say that they mimicked the ability of insects to monitor the passing landscape to be able to maneuver around obstacles as they fly, something referred to as optic flow. The nearest you can come to experiencing it is when you drive along a motorway: when you look forward, your view in front is quite stable, but if you look out to either side of your car, the side landscape seems to pass by much faster and faster. This reaches a maximum at a 90-degree angle to the path of your car.
The scientists equipped BeeRotor with just 24 photodiodes/pixels (which are distributed at the top and bottom of its eye) that it uses to measure optic flow. Despite their small number, the robot was able to use these pixels to detect contrasts in its flying environment and also their motion. Just like in insects, the angular velocity of the optic flow is found by measuring the speed at which features in the robot’s scenery move from one pixel to the next. When the optic flow increases, it means that the speed of the robot is also rising or that its distance relative to the obstacles is reducing.
BeeRotor has also been equipped with three feedback loops that act as its “brain” using three different reflexes that directly use the optic flow to make flight judgments. The first feedback loop is used to judge and change altitude, enabling BeeRotor to either follow the roof or the floor; the second loop is used to control the speed to adapt the robot to the size of the flight tunnel while the third one uses a dedicated motor to make stabilization adjustments to the eye relative to the local slope. Using these aspects, BeeRotor is able to always have the best field of view, independent of its degree of pitch. It can thus successfully avoid steeply sloping obstacles without the need for an accelerometer or measuring its speed or altitude. In late 2013, a patent had been taken out on this particular technology.
Owing to the success of the BeeRotor experiment, a new, biologically-plausible hypothesis could be put forward to explain how insects are able to fly without using an accelerometer: using feedback loops reminiscent of those that BeeRotor uses to fly, winged insects could be using cues obtained from optic flow to maintain their stability in the air.
Inertial reference systems that contain accelerometers are too bulky for small robots like BeeRotor. They have a mass of around one gram, something that makes them quite unsuitable for robots that weigh about 10 grams that could be applied, for example, in the inspection of piping. Lightness is also vital in the space industry, since each kilogram going into space comes with a significant cost. Even if optic flow sensors do not necessarily replace accelerometers entirely, they could, in case of failure on a space expedition mission, come in handy as an ultra-light back-up system. Check out this video demonstration of BeeRotor.
Image screenshot taken from Youtube (Video BeeRotor IROS 2012)