Brain-Inspired Technology Could Help Wearable Devices and Robots Use Less Power and Remember What They Have Detected
When a smartphone recognizes a face, one component captures the image, another stores the information and a processor interprets it. Most electronic vision systems including phone cameras to robots, work in this way. Signals must move between these separate components before the system can respond, adding delay, wiring and energy consumption.
A research team led by Khalifa University has developed a flexible device, known as a photomemristor that combines light sensing and memory in a single component and begins processing the light signal through changes in its electrical conductance.
This approach, known as in-sensor computing, follows a brain-inspired, or neuromorphic model in which adjustable electrical states perform functions similar to biological synapses.
The study was conducted by Dr. Shoaib Anwer, Dr. Muhammad Umair Khan, Nada AbuHamra, Prof. Mahmoud Al-Qutayri and Prof. Baker Mohammad from Khalifa University’s Centre for Intelligent Computing, Networks and Sensing (ICONS) and Department of Computer and Information Engineering, with Dr. Waqas Waheed from New York University Abu Dhabi.
The paper, ‘2D MXene-Enabled Unidirectional Flexible Optoelectronic Memristor Toward Energy-Efficient Neuromorphic Photosensing,’ was published in Energy & Environmental Materials, a top 1% journal.
Flexible and transparent, the device can fit onto curved surfaces where rigid electronics are difficult to use. By integrating sensing, memory, and early processing, the approach could reduce the energy and time required to move information between components, opening new possibilities for lighter, more responsive artificial vision systems.
By combining materials design, device engineering, and AI hardware evaluation, the project establishes a promising pathway from laboratory research toward practical technology.
The next steps include optimizing and demonstrating uniform, reliable device arrays; integrating these devices with electronics that read and process their signals; and validating the complete system for specific applications in ‘in-sensor’ computing and neuromorphic vision platforms.
Dr. Shoaib Anwer said: “By engineering the interfaces of 2D materials, we created a flexible device in which the 2D/2D material stack can detect light, retain the optical history, and provide analog states for neuromorphic computing with a unit cell. This combination is important for future in-sensor vision computing.”
Dr. Muhammad Umair Khan said: “The advance comes from device engineering and harvesting light-sensitive functional materials properties on an indium tin oxide-coated flexible polymer. Electrical testing showed self-rectifying, unidirectional analog switching with a rectification ratio of about ~103. Light provided a second way to program the same device to tune the memory.”
Alisha Roy
Science Writer
