Scientists Create an Artificial Eye That Could Give AI Human-Like Sight

TOKYO — Every second, your eyes process an enormous flood of visual information, yet your brain never crashes or overheats. Japanese researchers took inspiration from our eyes, creating an artificial retina that could give today’s power-hungry machine vision systems a major upgrade.

How the Artificial Eye Mimics Human Vision

The innovation, published in Scientific Reports, centers on what scientists call an “optoelectronic artificial synapse.” This is essentially a synthetic brain cell that responds to light. Just as your eye sends different electrical signals to your brain when you see red versus blue, this device generates positive or negative electrical responses depending on the wavelength of light it encounters.

Today’s machine vision systems capture and store every detail at 10 to 60 frames per second, which demands enormous computational resources. Your eyes, by contrast, are remarkably selective. They filter and compress information before sending it to your brain, making human vision incredibly energy-efficient. The researchers say modern systems struggle to work well in small, low-power devices because they need too much data and storage to function.

The research team from the Tokyo University of Science tackled this challenge by creating a device that combines two dye-sensitized solar cells. These act like tiny solar panels that have been modified to respond differently to various colors of light. One cell is designed to respond strongly to blue light, while the other prefers red. When light hits the device, it generates either a positive voltage for blue wavelengths or a negative voltage for red ones.

This bipolar response, the ability to produce both positive and negative signals, is what makes the device so powerful. Traditional photodetectors can only produce positive signals, limiting their ability to distinguish between different types of input. But this new system can tell the difference between wavelengths as close as 10 nanometers apart, giving it color discrimination abilities that rival human vision.

To test their creation, the researchers demonstrated that the device could perform basic logic operations, the building blocks of all computing. Unlike conventional devices that struggle with complex operations, this artificial eye handled these functions with ease, thanks to its ability to switch between positive and negative responses based on light color and intensity.

But the real test came when they used the device for what’s called “physical reservoir computing,” a brain-inspired approach to machine learning where the physical properties of a material do the heavy lifting instead of traditional computer processors. The team flashed sequences of colored light at the device, each representing a different code. It was able to tell apart patterns up to six digits long, meaning it could recognize 64 different combinations. That’s impressive for such a simple setup.

The researchers tested the device’s ability to recognize human motion in different colors. They recorded videos of six different actions—bending, waving with one hand, waving with both hands, jumping, running, and moving sideways—and converted them into sequences of red, green, and blue light pulses. The artificial eye achieved 82% accuracy in identifying both the motion and the color, a performance that approaches what you might expect from much more complex systems.

How Is It So Efficient?

The device is based on solar cell technology, meaning it generates its own electricity from the light it’s detecting. This eliminates the need for external power sources and could enable a new generation of autonomous sensors that can operate indefinitely in natural lighting conditions.

While conventional machine vision systems require constant power to process and store massive amounts of visual data, this bio-inspired design filters out extra information right in the hardware, so it doesn’t need nearly as much computing power to work.

Scientists envision applications ranging from compact surveillance systems to medical devices that monitor vital signs using multiple wavelengths of light. The technology could be particularly valuable for “edge computing” applications, situations where devices need to operate independently without connection to powerful central computers. Potential examples could include security cameras in remote locations, environmental sensors in forests, or health monitors that patients wear continuously.

The researchers do mention that these possibilities would require more sophisticated signal processing to handle complex visual scenes. However, they believe these challenges can be addressed by optimizing the dye materials and device architecture.

Every device around us is constantly growing hungrier for power and processing. While we’re still years away from artificial eyes that fully match human vision, but this work takes a step toward artificial vision systems that combine the efficiency of biological sight with the precision of electronic devices.