Researchers have developed tiny devices that see and remember in the same way humans do, a promising step toward one day having applications that can make fast and complicated inferences, such as self-driving cars.
The neuromorphic device is a single chip powered by a doped indium oxide sensing element that is hundreds of times thinner than a human hair and requires no other components to function. Engineers from RMIT University in Australia led the work with input from researchers from Deakin University and the University of Melbourne.
The team’s research demonstrates a working device that captures, processes and stores visual information. With a precise design of doped indium oxide, the device mimics the human eye’s ability to capture light, prepackages and transmits information like an optic nerve, and stores and classifies it in a memory system like our brain can.
Together, these features could enable ultra-fast decision-making, the team says. Team leader Professor Sumeet Walia said the new device can perform all the necessary functions – sensing, creating and processing information, and storing memories – rather than relying on external power-hungry computations that hinder real-time decision-making.
“Performing all these functions on one small device has so far proven to be a big challenge,” said Walia from RMIT’s School of Engineering.
“With our invention, we have made real-time decision making possible because it does not need to process large amounts of irrelevant data and is not slowed down by transferring data to separate processors.”
The new device was able to demonstrate the ability to retain information for longer periods of time, compared to previously reported devices, without the need for frequent electrical signals to refresh the memory. This capability significantly reduces power consumption and increases device performance.
Their findings and analyzes are published in Advanced Functional Materials.
First author and RMIT PhD researcher Aishani Mazumder said the human brain uses analog processing that allows it to process information quickly and efficiently with minimal energy.
“In contrast, digital processing is energy and carbon intensive and prevents the rapid collection and processing of information,” she said.
“Neuromorphic vision systems are designed to use analog processing similar to the human brain, which can significantly reduce the amount of energy needed to perform complex visual tasks compared to today’s technologies.
The team used ultraviolet light as part of their experiments and is working to further extend the technology to visible and infrared light – with many potential applications such as bionic vision, autonomous operations in hazardous environments, food shelf life assessment and advanced forensics. .
“Imagine a self-driving car that can see and recognize objects on the road in the same way a human driver can or is able to quickly detect and track space junk. This would be possible with neuromorphic vision technology.”
Walia said neuromorphic systems could adapt to new situations over time and become more efficient with more experience.
“Traditional computer vision systems – which cannot be miniaturized like neuromorphic technologies – are usually programmed according to specific rules and cannot adapt as easily,” he said.
“Neuromorphic robots have the potential to work autonomously for long periods of time in hazardous situations where workers are exposed to possible falls, explosions and toxic air.”
The human eye has a single retina that captures the entire image, which the brain then processes to identify objects, colors, and other visual elements.
The team’s device mimicked the capabilities of the retina using single-element image sensors that capture, store and process visual information on a single platform, Walia said.
“The human eye is uniquely capable of responding to changes in the surrounding environment faster and much more efficiently than cameras and computers currently can,” he said.
“We were inspired by the eye and have been working for several years to create a camera that has similar capabilities through a process of neuromorphic engineering.”
