An example of an AI-designed unidirectional imager that can only produce images in one direction and that can function regardless of the polarization of the light across a wide range of wavelengths. Credit: UCLA’s Ozcan Lab
Conventional optical imaging and communication systems, which are usually made up of lenses, may image both forward and backward. Standard optical materials and lenses allow light to travel both ways via devices like a camera, much like a pipe that allows liquid to flow through from one end to the other in both directions.
In order to challenge this paradigm in optical imaging, a research team at the UCLA Samueli School of Engineering has developed a new unidirectional material. In order to structurally build materials with details at sizes smaller than the wavelength of light, the researchers employed artificial intelligence (AI). As a result, the imager design only allows for imaging in one direction while blocking off the other.
The publication “Unidirectional Imaging Using Deep Learning-Designed Materials” was published today in Science Advances. It describes a novel optical imager design that uses deep learning to spatially engineer materials into a sequence of transmissive optical layers.
“This AI-driven engineering of the material leads to the formation of multiple structured, transparent layers that collectively modulate the optical waves with an asymmetric behavior in the forward and backward imaging directions,” stated Aydogan Ozcan, the leading investigator of the study and the Volgenau Chair for Engineering Innovation at UCLA. Ozcan is a Samueli professor of electrical and computer engineering at UCLA. “Image generation from an input field of vision to an output field of view is possible with the resulting unidirectional imager, but not the other way around. The picture formation’s reverse path is inhibited.”
Despite being trained with a single illumination wavelength, the unidirectional imager can continue to perform under broadband light that consists of many wavelengths. Additionally, this unidirectional imaging ability operates regardless of the direction of light oscillations and is unaffected by light polarization. The researchers used terahertz radiation to effectively show the design’s effectiveness on a 3D-printed multilayered imager in their studies.
The group also created a different imager that let the user select a wavelength to determine which direction the image should be blocked. For instance, at one wavelength, the picture generation is limited to the left, whereas at another wavelength, the image can only be seen from the reversed path.
The unidirectional imager’s versatility and capability are increased by this wavelength-multiplexed design, which enables it to operate like a switchboard to regulate the flow of data encoded in light waves. Unidirectional imagers are extremely thin, having a thickness of a few tens of wavelengths, which in the visible spectrum would be equivalent to the thickness of a stamp. They can work at any point in the electromagnetic spectrum utilizing various transmissive materials or substrates. These unidirectional imagers may have a big impact on defense, security, and telecommunications, among other industries.