The idea of the invisibility cloak is one of the most anticipated technologies in recent years, mostly due to the popularity of Harry Potter. However, rather than magic, it’s becoming a reality thanks to Professor Chen Hongsheng of Zhejiang University, College of Information Science and Electronic Engineering.
His proposed cloaking mechanism uses a pre-trained artificial neural network (ANN). This feature gives the cloak self-adaptation without the need for human intervention.
Closer to invisibility
The invisibility cloak research was published under the journal Nature Photonics last March 23. Here, it describes how it applies transformation optics, where the illusion of “invisibility” comes from the guiding of light around an object.
Rapid and automatic internal structure adjustment, along with the ability to retain its invisibility with the application of external stimuli or non-stationary environment, make up the ideal invisibility cloak. Although desirable, these characteristics are not easily accessible since there are various other factors to be developed for future applications.
Chao Qian, the lead author of the study, cited the octopus as the “master of disguise”. This is due to its quick response and physical alteration from the chromatophores, iridophores, and leucophores present in its skin. He stated how their research could employ a similar process, saying “we can incorporate tunable metamaterials, deep learning, and EM waves in the design of the cloak.”
Notable advances in transformation optics and synthetic electromagnetic (EM) materials have helped research on invisibility methods push forward. For the past three years, the team attempted to develop an intelligent invisibility cloak. This was not an easy feat considering that the team had to begin from square one to create an intelligent invisibility system.
How invisibility works
Each element inside the metasurface has a reflection property. This could be independently tuned by applying various direct-current bias voltages at microwave frequencies across a loaded varactor diode. With the application of ANN, the metasurface cloak can respond to the dynamic incident wave on a millisecond timescale.
The bias voltages are automatically calculated and fed instantaneously to the cloak. Using a finite-difference time-domain (FDTD) program, the pre-trained ANN can copy a real-world display. The researchers then employed a proof-of-concept to benchmark the performance.
In comparison, the ability of the metasurface cloak is better than an actual chameleon, considering that it takes around 6 seconds for the animal to blend with its background. Meanwhile, the cloak takes only approximately 15 milliseconds for it to offset external lighting and complete the camouflage.
Optical ANNs, equipped with nanophotonic circuits and metasurfaces, bring forth the prospect of enhanced computing speed and power efficiency. With this innovation, existing clothing designs may be altered for use in practical scenarios with various backgrounds and moving objects, not to mention, it could be immediately scaled up to higher frequencies. Future meta devices may develop from the concept, and may also contribute to solving various other issues such as the EM inverse problem.
Photo credits: The feature image was taken by Dong Jiahui. All images displayed are owned by Zhejiang University and were provided for press usage.
Source: ZJU press release
