Scientists have developed an object detection camera that mimics the structure and capabilities of an eagle’s eye. (Image courtesy of IBS)
DAEJEON, May 31 (Korea Bizwire) – In an emulation of the sky’s ultimate predator, scientists have developed an object detection camera that mimics the structure and capabilities of an eagle’s eye, including the ability to perceive ultraviolet light.
The Center for Multidimensional Carbon Materials at the Institute for Basic Science (IBS) announced on May 30 that a joint research team led by associate director Kim Dae-Hyeong and professor Song Young-Min from the Gwangju Institute of Science and Technology (GIST) has created a camera inspired by the avian eye.
The eyes of eagles have evolved to have a deep and narrow fovea (the center of the retina) to adapt to living high up in trees. This eye structure allows them to spot moving prey from kilometers away.
Crucially, the fovea in birds’ eyes, including eagles, is densely packed with cone cells that detect color, enabling sharper perception of objects. Unlike the human eye limited to visible light, avian eyes can also detect ultraviolet radiation, providing access to visual information invisible to humans.
The research team designed an artificial fovea that mimics the structure and function of a bird’s eye, creating a camera specialized for object detection. They integrated a multi-spectral image sensor capable of detecting not only visible light but also ultraviolet radiation.
Utilizing perovskites (oxides exhibiting properties of insulators, semiconductors, conductors, and even superconductors) renowned for their excellent electrical and optical characteristics, the researchers fabricated light sensors from four different perovskite materials that absorb distinct wavelength ranges.
By stacking these sensors, they achieved a color-discriminating sensor without the need for color filters.
Unlike conventional zoom cameras that magnify objects but struggle with peripheral vision, this object detection-specialized camera offers a wide field of view, capturing both central and peripheral areas.
The researchers explained that the camera’s motion detection rate is 3.6 times higher than existing camera systems, enabling agile movement tracking. Additionally, its confidence score, a measure of object recognition ability, is approximately twice as high as conventional cameras.
“By eliminating filters, we can reduce production costs and weight,” said Kim. “This technology can be applied to unmanned robots, autonomous vehicles, and is expected to be particularly effective for drones operating in environments similar to birds.”
The research findings were published in the online edition of the journal Science Robotics on May 30.
Kevin Lee (kevinlee@koreabizwire.com)
