Donate to Science & Enterprise

S&E on Mastodon

S&E on LinkedIn

S&E on Flipboard

Please share Science & Enterprise

Sensor Designed for A.I. Mimics Eye Functions

Red-blue-green eye

(Public Domain Pictures, Pixabay)

9 Dec. 2020. A computer science lab designed an optical sensor to function more like a human eye and better process visual data for systems based on artificial intelligence. The team of engineering and computer science professor John Labram and graduate student Cinthya Trujillo Herrera at Oregon State University in Corvallis describes the sensor in yesterday’s issue of the journal Applied Physics Letters.

Labram and Trujillo are investigating better and simpler sensors for applications using vision analysis, such as image recognition, self-driving vehicles, and other robotics. Current optical sensors rely on complex circuitry or software to emulate the capabilities of the human retina. The retina is a layer on the back of the eye with light-sensitive cells that convert light into neural signals, carried into the brain by the optic nerve.

At the same time, say the researchers, current optical sensors process their data as a series of commands for conventional computer processing, rather than for systems built around neural networks, a form of artificial intelligence that process often massive quantities of data in parallel. Labram notes in an university statement that, “even though the algorithms and architecture designed to process information are becoming more and more like a human brain, the information these systems receive is still decidedly designed for traditional computers.”

The Oregon State team designed its sensor to function more like the human retina that pre-processes light waves with photoreceptors before transmitting electrical signals to the optic nerve. The sensor is built with ultra-thin layers of perovskite, a low-cost calcium titanium oxide material used increasingly in solar cells for their light absorption and electrical conductivity. The perovskite layers are tuned to act as capacitors for rapid charging and discharging, built on a silicon dioxide surface with thermally-deposited electrodes.

The optical sensor responds to light by transmitting a temporary spike in current before returning to its previous state. Constant illumination is processed as a constant low-voltage current, while changes in illumination, as when visualizing motion, create higher voltages for transmission and statistical reconstruction in a computer vision application. The researchers demonstrated the optical sensor in simulations, with data from their prototype sensor combined to represent capturing motion in A.I.-enabled systems.

“For example,” says Labram, “you can imagine these sensors being used by a robot tracking the motion of objects. Anything static in its field of view would not elicit a response, however a moving object would be registering a high voltage. This would tell the robot immediately where the object was, without any complex image processing.”

Labram adds, “The good thing is that, with this simulation, we can input any video into one of these arrays and process that information in essentially the same way the human eye would.”

More from Science & Enterprise:

*     *     *

Comments are closed.