Shrimp Vision to Aid Driverless Cars

Crustaceans are a bit like the superheroes of the underwater world. With their rugged exoskeletons and menacing claws, they look straight out of a fantasy flick. Some of them, like the lobster, can easily live more than a century. And, although they are usually coveted for their culinary qualities, this time they have inspired a cutting-edge vision technology with great potential for driverless cars. This small crustacean, which lives in the Australian coral reefs goes by the name of mantis shrimp (gonodactylus smithii), has several outstanding skills. Firstly, it punches literally above its weight as it can strike with the speed of a bullet (350 meters per second). And, more relevantly to the topic at hand, it makes use of an extremely complex visual system that allows it to detect its prey in shallow waters with high contrasts. And those high contrast situations pose a real technological challenge for driverless cars.

In 2016, the first death in an accident involving a driverless car was recorded in the USA. The reason? The onboard cameras were unable to distinguish a sunny sky from a white lorry ahead. The incident drove Viktor Gruevm a researcher from the University of Illinois, to wonder if there could be a technological solution for the issue. And so, he started studying the visual system of the mantis shrimp, able to see detailed nuances in extremes of light and dark at the same time. This is technically called high dynamic range and allows some of the latest TVs, like those featuring OLED panels, to provide high luminosity and deep blacks simultaneously. The mantis shrimp detects polarization, a property of light where light waves that vibrate in more than one plane are restricted to vibrate along a single plane. This means that its visual system has a logarithmic response to light instead of a linear one.

Gruev replicated this logarithmic response in new camera technology with half a million sensors that detect the dark and bright areas of a scene. Thus, his cameras offer ten thousand times more contrast than current systems, which means they can distinguish objects regardless of their color and work even in hazy conditions. Another advantage of this technology is that it’s based on commercially available components, so this type of devices should hit the market soon. Finally, these cameras could also have interesting applications in the detection of cancerous cells and underwater exploration.

How do driverless cars “see”?

Currently, driverless cars use several technologies to assess their environment. In addition to traditional cameras, they are equipped with LIDAR systems that employ a pulsed laser to establish the distance of the objects in the surroundings. This technological innovation, however, is color-blind, so other systems, such as thermal imaging, need to be added to the mix. All this information is then fed into an AI system to interpret the context, discriminating between humans, animals, vehicles or obstacles among other things. Driving under real conditions, with millions of variables involved, is an undeniable challenge that will require the most advanced technologies. And, as the mantis shrimp proves, nature can lend us a helping hand.

Source:  Engineering.com