Researchers at RMIT University have developed a prototype vision system that can sense, store and process information on the same device, with the technology potentially paving the way for smart bionic eyes and lower-energy AI systems.
The neuromorphic vision system is designed to mimic aspects of the way the human eye and brain work together, reducing the need to transfer large amounts of data between separate sensors, memory and processors.
The early-stage prototype uses the atom-thin semiconductor molybdenum disulfide (MoS₂), with sensing, processing and storage taking place on a 2cm by 2cm chip.
Researchers have trained the system to recognise numbers, shapes and movement. In laboratory testing, it was able to detect changes in its field of view, store the information as memory and process it locally.
RMIT Centre for Opto-electronic Materials and Sensors (COMAS) Professor Sumeet Walia said the work was inspired by the efficiency of biological vision.
“The human eye and brain work together incredibly efficiently, processing vast amounts of information using remarkably little energy,” Prof Walia said.
“Our research is helping lay the foundations for technologies that work in a more similar way.”
Unlike a conventional camera, which captures frames before sending the information elsewhere for processing, the RMIT system performs much of that processing where the information is collected.
Co-researcher Dr Taimur Ahmed said this reduced the need to continually move information between different parts of a computing system.
“This is not just a sensor that captures information, it’s a sensor that can also process information,” Dr Ahmed said.
“Rather than constantly moving data between separate memory and processing units, much of that work happens much closer to where the information is generated.”
The researchers said the approach could eventually be used in a smart bionic eye capable of identifying changes in a scene, storing relevant information and rapidly processing visual signals while consuming relatively little energy.
The team has also developed a water-based manufacturing process for transferring atom-thin semiconductors and electrodes, which it said produced fewer defects and improved electrical and light-sensing performance compared with conventional methods.
“Each breakthrough brings us closer to technologies such as a smart bionic eye,” Prof Walia said.
While practical applications remain years away, the researchers said the technology could also have applications in robotics, autonomous vehicles, machine vision and intelligent sensors.
It could also provide a lower-energy approach to some AI applications by processing information closer to where it is collected, reducing the amount of data that needs to be transmitted, stored and analysed.
“If this RMIT technology can be scaled up, it could help reduce the amount of data that needs to be moved, stored and processed, making future AI systems more energy efficient,” Prof Walia said.
RMIT has filed an international patent application for the technology under the Patent Cooperation Treaty.
Add Comment