A new shrimp-inspired camera could help autonomous cars see better

Have you ever looked into an animal's eyes and thought, "I wonder how they see the world?" We did it. We do it all the time. We wonder, for example, what our cat sees when she walks in the woods and how the colorful world appears to a color-blind dog. And when a cockroach crosses the kitchen counter late at night, we wonder if he can feel the disgust in our unfocused look.

We know that we are not the only ones thinking these things. Viktor Gruev, electrical engineer and computer scientist at the University of Illinois at Urbana-Champaign, shares our plot – although from a more technical point of view.

Gruev and his colleagues have recently developed a camera prototype inspired by mantis shrimp, which has one of the most complex visual systems in the animal world. By imitating the vision of the marine crustacean, researchers have been able to significantly improve current commercial cameras and could help mitigate accidents by allowing autonomous vehicles to see more clearly.

"We introduced a new camera prototype inspired by the mantis shrimp visual system, which allowed us to capture polarization information in a high dynamic range," Gruev told Digital Trends. "This is important because the polarization of light is a fundamental property of light that we [as humans] are blind to. It provides essential information about the environment represented, such as the material properties of objects and their geometric shape. It is also useful to capture this information when driving in hazy or foggy conditions, where polarization information can improve the detection visibility of objects on the road. "

As for the dynamic range, it is a measure of the lightest and darkest areas that a camera can capture at a given moment. Imagine driving in a dark tunnel and going out on a sunny day. For a few seconds, the sunlight is almost blinding. Increased dynamic range allows cameras to more easily adapt to this transition.

Inspired by recent car accidents driving self-driving, Gruev decided to tackle the problem of pedestrian detection through biomimicry. He and his team have adopted the traits of mantis shrimp, able to simultaneously see bright and dark images.

"There are two innovations in this work," said Gruev. "The pixels and nanomaterials we use for polarization detection. The pixel is new because it works very differently from conventional cameras, which allows a logarithmic response to light intensity. This change allowed us to extend the dynamic range of the camera by approximately 60 dB, which is typical of conventional cameras, at around 140 dB. We then added aluminum nanowires on each pixel of the imager to be able to detect the polarization properties of the light. "

Gruev plans to use his camera to support autonomous cars and an underwater GPS. He said that they are currently working with automakers to commercialize the technology.

An article detailing the research was recently published in the journal Optica.