Metasurface smart glass for fast and secure object recognition

Unmet Need: Materials for efficient and secure object recognition systems

The increasing use of computer vision features on mobile devices, cars, and other technologies has created a demand for efficient and secure object recognition systems. Artificial neural networks (ANNs) can be used for this purpose, but they require significant power and may have latency and security vulnerabilities. Optical neural networks (ONNs) using metamaterials could be a better solution, but the difficulty of fabricating 3D metamaterials is currently limiting their use.

The Technology: Optical neural network comprised of 2D metasurfaces

This technology is a “smart glass” optical neural network composed of layered 2D metasurfaces. These surfaces are comprised of arrays of variously shaped 2D nanopillars, which can manipulate incoming wave properties. This passive device handles complex, high-throughput tasks at the speed of light, using little to no power. Since the object is not represented within the domain, data security is guaranteed.