(Nanowerk Spotlight) Hand gestures have become a natural way to communicate with digital systems, but making machines understand those gestures with precision remains a technical challenge. While cameras and wearable devices have been used to translate hand movements into commands, both solutions come with trade-offs. Optical systems often falter in poor lighting or cluttered environments, while wearable devices can feel invasive and uncomfortable. As technology advances, the need for a more seamless, reliable way to interpret human gestures has grown more pressing.
A new approach, however, could change the game. Researchers are turning to programmable topological metasurfaces—thin, engineered materials that can manipulate electromagnetic waves with great precision. These metasurfaces allow machines to sense gestures wirelessly by detecting the subtle ways in which hands interfere with electromagnetic fields. Unlike camera-based systems, which rely on light, or wearables, which track physical motion, this method captures gestures through invisible electromagnetic changes, offering a potentially more reliable and unobtrusive alternative.
In a study published in Advanced Functional Materials (“Intelligent Hand-Gesture Recognition Based on Programmable Topological Metasurfaces”), a team of researchers from Southeast University and the City University of Hong Kong presented a significant advancement in this area using programmable topological metasurfaces. Their research introduces a robust and highly accurate hand-gesture recognition system based on these advanced materials, overcoming many of the limitations of previous systems. By leveraging the unique properties of topological metasurfaces – thin layers of material designed to manipulate electromagnetic waves in specific ways – they were able to create a system that reliably recognizes both simple and complex hand gestures without the need for cameras or wearables.
The schematic of the hand-gesture recognition system based on programmable topological metasurfaces. The programmable topological metasurface controlled by FPGA switches five propagation paths to ports 2–6 for transmission coefficient collection using VNA. Based on these multidimensional EM data, a well-trained neural network can accurately classify five single-hand gestures and 25 two-hand gestures. (Image: reproduced with permission by Wiley-VCH Verlag)
The core of this new approach lies in the metasurfaces themselves, which are engineered to control surface waves—electromagnetic waves that travel along the surface of the material. These surface waves are sensitive to objects that come close to them, such as a human hand, and can be dynamically programmed to interact with those objects in specific ways. In this system, the metasurface generates surface waves that are altered by the presence of a hand, and those alterations can be measured to detect different hand gestures.
One of the key innovations in this study is the use of programmable topological metasurfaces, which can change their configuration on the fly. The researchers embedded the metasurface with PIN diodes—electronic switches that can be turned on and off to control how electromagnetic waves propagate across the surface. By using a field-programmable gate array (FPGA), a device that can quickly adjust these configurations, the system can switch between multiple propagation paths in real-time. This flexibility allows the system to capture more detailed information about hand gestures by sampling electromagnetic data from different angles and paths.
The researchers developed a setup where the metasurface interacts with a vector network analyzer, a tool used to measure electromagnetic wave transmission. When a hand is placed above the metasurface, it disturbs the electromagnetic waves traveling across it. These disturbances, called transmission parameters, vary depending on the specific hand gesture. The system records these changes and processes them through a neural network that has been trained to recognize different gestures. The neural network, which is a type of machine learning algorithm, learns to classify gestures based on the unique electromagnetic signatures they produce.
In their experiments, the researchers tested five single-hand gestures, such as a fist, thumb-up, or open hand, as well as 25 combinations of two-hand gestures. They collected more than 5,000 sets of transmission data, using this to train the neural network. The results were impressive: the system was able to recognize individual hand gestures with an accuracy of over 99%. For two-hand gestures, the system achieved similarly high performance, with 100% accuracy in some cases.
What sets this approach apart from previous methods is not just the accuracy, but also the robustness of the system in real-world conditions. Traditional electromagnetic sensing systems often struggle with external interference – background noise from other signals in the environment can distort the measurements, leading to lower accuracy. However, the use of topological metasurfaces provides a level of protection against such interference.
Topological materials are known for their ability to maintain stable wave propagation, even in the presence of defects or external disturbances. This means the system can function reliably even in environments that would be challenging for other technologies, such as crowded urban areas or industrial settings where multiple electronic devices are in use.
The neural network used in the system is designed with three hidden layers and 100, 50, and 20 neurons, respectively. It processes the electromagnetic data collected by the metasurface and learns to recognize distinct patterns associated with each hand gesture. One of the most important aspects of the neural network’s design is its ability to generalize well to new data. In other words, even when tested with hand gestures it had not seen before, the system maintained its high accuracy. This level of reliability is crucial for practical applications, where gestures may vary slightly from person to person or from one instance to the next.
An interesting part of the research was the exploration of how factors like the height of the hand above the metasurface or the frequency range of the electromagnetic waves affected the system’s performance. The researchers found that the system worked best when the hand was within 5 to 15 centimeters of the surface. Beyond this range, the electromagnetic coupling – the interaction between the hand and the surface waves – became too weak to provide accurate data. Similarly, the number of frequency points used in the measurement also affected accuracy. The more frequency points the system measured, the better it performed, with a significant drop in accuracy when fewer than 10 frequency points were used.
While the current system relies on a vector network analyzer, which is a sophisticated and relatively expensive piece of equipment, the researchers believe that future versions could be made more compact and affordable. One possibility is to replace the VNA with simpler spectrum sensors combined with high-quality filters. This would make the technology more accessible for commercial applications, such as smart homes, gaming, or medical devices, where the ability to control systems with simple hand gestures could offer significant benefits.
The potential applications for this technology are vast. In environments where hygiene is critical, such as hospitals or food processing plants, touchless control systems could reduce the risk of contamination. In virtual reality, more precise and reliable gesture recognition could improve the user experience by making interactions more natural and immersive. Even in daily life, this system could enable new forms of interaction with smart devices, eliminating the need for remote controls or touchscreens. The combination of high accuracy, robustness to interference, and flexibility makes programmable metasurfaces a promising platform for the next generation of gesture recognition systems.
Get our Nanotechnology Spotlight updates to your inbox!
Thank you!
You have successfully joined our subscriber list.
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com.
