“Beauty is in the eye of the beholder,” said Plato, or something along those lines with a more Grecian flare. Beauty is however distinctly affected by the kind of eye beholding it – compound or single lens. The human eye comprises a single lens through which light streams in and is detected by light-sensing structures at the back of the eye. This set-up allows us to see far into the distance and with great resolution. Arthopods, like bees and flies, however sport compound eyes. These comprise hundreds or thousands of light-sensing structures that are densely packed onto the curved surface of the eye.
Although compound eyes lack such good resolution, they are able to detect very fast movements due to the panoramic vision afforded to them by each light-sensing unit pointing in a slightly different direction. These traits make the compound eye appealing to aerospace engineers striving to improve wide field motion detection for improved navigation of vehicles and collision avoidance.
European researchers teamed up to tackle the challenge of making an artificial compound eye. The result was a three-layered, flexible device called CurvACE, capable of capturing panoramic imagery without distortion. CurvACE packs all of this punch into a hemispherical block no bigger than a cherry and uses 0.9 Watts at maximum power (for comparison: it takes about 5 Watts to fully charge an iPhone). Taking inspiration from nature, the design of CurvACE was based on the fruit fly whose eyes each contain hundreds of light-sensing units called ommatidia (a self-explanatory term for old Plato, omma being Ancient Greek for eye).
Just like the fruit fly, the first of the three layers of CurvACE, captures and focuses light. This layer comprises 630 microlenses made of a transparent polymer. The focused light then hits a layer of photodetectors. This array of light-sensitive silicon detectors, each aligned with a single microlens, converts the light into a comprehendible signal for the third and final layer – the printed circuit board. The circuit board relays these signals to a computer for processing. Each layer when sandwiched together produces a flexible structure just a millimetre thick. The assembly of 630 artificial ‘ommatidia’ was carefully bent into a 180°-curved panel. The space behind this hemispherical surface was filled with the electronics needed to smoothly control the movements of CurvACE and report the data collected.
With the prototype finally complete, the researchers set about determining the visual limits of CurvACE. Firstly, testing each of the individual ommatidia demonstrated that CurvACE could successfully sample a panoramic field of view stretching across 180°. To make the tests even harder the researchers played with the ambient light. CurvACE proved sensitive enough to function in low light conditions but was also able to adapt to bright light conditions, without overloading its sensors. The final trial looked at how CurvACE would fair with motion detection.
Implementing a range of motions using a black and white patterned wall, the researchers showed that the visual signals detected by CurvACE accurately detected motion in three different light conditions. Moreover, CurvACE one-upped your average insect when it came to the range of signals it could detect. In tech speak the ‘signal acquisition bandwidth’ of CurvACE was 300Hz - three times greater than ommatidia of fast-flying insects. The lower the bandwidth the more visual distortions occur, especially during fast movements. Something of particular importance to a fly making a speedy retreat from a swatter or to future iterations of CurvACE trying to prevent aerospace collisions.
CurvACE marks a significant step in the development of biomimetic eyes. Being a prototype there is much left to tinker with – increasing the resolution of images acquired, decreasing its size and expanding the field of view to name but a few. Teams of biorobotics researchers are on the case, so keep your very human eyes peeled.
Floreano D, et al. (2013) Miniature curved artificial compound eyes. Proceedings of the National Academy of Sciences, vol. 110, no. 23, pp 9267-9272.