Scientists from the School of Engineering and Applied Sciences University of Virginia have developed innovative artificial eyes capable of overcoming the limitations of modern machine visual perception systems. The inspiration for this work was the visual system of praying mantises, insects that have a unique ability for binocular vision.
Scientists explained: traditional systems of visual perception of unmanned vehicles and drones face problems of processing static or slowly moving objects in three-dimensional space. Many insects with compound eyes have a good ability to track movement, but are poor in depth perception. The exception is praying mantises, in which the field of vision crosses between the left and right eye, creating binocular vision and depth perception.
Inspired by this phenomenon, scientists created artificial compound eyes by combining optoelectronic engineering and innovative “edge” technologies. computing. This approach allows processing data directly in the sensors or next to them. In turn, this significantly reduces processing time and computing power requirements.
Dr. Byeongjun Bae, a graduate student in the Department of Electrical and Computer Engineering, explained that creating a biomimetic system that replicates the biological abilities of mantises requires new technologies. The team used microlenses and photodiodes, which generate an electric current under the action of light, and also flexible semiconductor materials to simulate the convex shapes and position of facets in mantis eyes.
The progressive system provides accurate spatial perception in real time, which is especially important for use in dynamic environments — drones, robotic production line systems, as well as surveillance and security systems, "smart" home devices. One of the team's significant achievements was reducing power consumption by more than 400 times compared to traditional visual systems.
Professor Kyusan Lee, Bae's mentor, said their work could inspire other engineers and scientists to develop biomimetic solutions for complex tasks of visual processing. This technology opens up new possibilities for the use of artificial eyes in various fields, from unmanned vehicles to robotics and smart security systems.