In a landmark development, scientists at the City University of Hong Kong have engineered a state-of-the-art wearable bionic device that remarkably emulates the extraordinary vision capabilities of birds while operating with minimal energy. This breakthrough in machine vision technology holds the potential to revolutionize autonomous systems by enhancing their interaction with the environment, according to a recent publication in Nature Communications.
Machine vision is integral to various applications where rapid object identification and classification are critical, such as in autonomous driving and robotics. However, conventional silicon-based vision chips are hampered by high energy demands and struggle to replicate the complex behaviors seen in biological systems. Addressing these challenges, the research team, led by Professor Johnny C. Ho, has fused advanced materials with neural network architectures, creating a vision system that boasts superior performance even under low-light conditions while maintaining energy efficiency. It also facilitates broadband non-volatile storage capabilities.
Professor Ho, who serves as Associate Vice-President (Enterprise) and is a professor in the Department of Materials Science and Engineering at CityUHK, highlighted the transformative potential of this research. “By integrating large-area wearable visual bionic devices onto any surface, AI hardware can be propelled into everyday use, merging seamlessly with sophisticated algorithms,” Professor Ho explained. “Employing arrays of neuromorphic devices in tandem with deep neural networks can greatly improve recognition efficiency and curtail energy usage.”
The team’s novel design hinges on specially aligned gallium arsenide (GaAs) nanowire arrays paired with liquid-surface-assembled P3HT organic films, creating van der Waals heterojunctions. This innovative system deploys reservoir computing to successfully detect and classify a multitude of features in moving objects, encompassing shape, motion, color, and ultraviolet (UV) grayscale attributes.
Achieving precise molecular orientation within semiconductor films was a significant breakthrough for the team, providing reliable and consistent performance across varying surfaces. This adaptability underpins the technology’s potential for a wide array of applications, from smart driving systems to robotic vision and beyond.
Looking ahead, the researchers aim to push the boundaries of this technology further by integrating it with external circuits, paving the way for seamless interaction between hardware and software. The development of intelligent visual processing systems might soon take a giant leap forward, reshaping the landscape of machine vision technologies.