Researchers at Tsinghua University have developed a new optoelectronic device that mimics the human visual system for diversified in-sensor computing. The device, detailed in Nature Nanotechnology, is based on a fully integrated array of optoelectronic memristors (OEMs), which are hardware components capable of both processing and storing information. The human visual system processes information hierarchically, from basic sensory stimuli to advanced cognitive abilities.
The researchers aim to replicate this hierarchical structure in systems to effectively handle various information processing levels. In-sensor computing, which integrates sensing, memory, and processing functions into a single device, is a promising approach to achieve this goal. The device developed by Heyi Huang, Xiangpeng Liang, and their colleagues features a fully integrated 1-kb array with 128 × 8 one-transistor one-optoelectronic memristor (OEM) cells and silicon CMOS circuits.
It has configurable multi-mode functionality, including electronic memristor, dynamic OEM, and non-volatile OEM modes. The OEMs have a layered structure with materials like Pd/TiO2/ZnO/TiN.
New optoelectronic in-sensor computing platform
The different configurable modes of operation allow the system to mimic the hierarchical information processing of the human visual system. Initial experiments have shown promising results. The device was used to run computer vision algorithms, achieving high accuracy on three visual tasks: image sensory pre-processing, object tracking, and human motion recognition.
The NV-OEM mode enhanced image sensory pre-processing accuracy from 85.7% to 96.1%, while dynamic OEM and NV-OEM modes achieved 96.1% accuracy in object tracking. The fully OEM-based in-sensor reservoir computing system achieved 91.2% accuracy in human motion recognition, consuming over 20 times less energy than traditional graphics processing units. This new optoelectronic platform offers cost-effective solutions for a variety of in-sensor computing applications.
Future research by the team may involve optimizing system performance, such as using transparent materials on the OEM’s top electrode to increase light absorption. For more information, see the full paper by Heyi Huang et al. in Nature Nanotechnology.
Rashan is a seasoned technology journalist and visionary leader serving as the Editor-in-Chief of DevX.com, a leading online publication focused on software development, programming languages, and emerging technologies. With his deep expertise in the tech industry and her passion for empowering developers, Rashan has transformed DevX.com into a vibrant hub of knowledge and innovation. Reach out to Rashan at [email protected]
