A smart contact lens capable of controlling robots with just eye movements has been developed in Korea. The technology is expected to serve as a next-generation ultra-lightweight extended reality (XR) interface that could replace existing heavy and complex XR devices.
A research team led by Professor Chung Im-doo of the Department of Mechanical Engineering at UNIST (Ulsan National Institute of Science and Technology), who also holds a joint appointment at the Graduate School of Artificial Intelligence, announced on the 15th that they have developed a smart contact lens capable of remotely controlling a robotic arm. The key innovation lies in combining specialized technology that directly prints extremely small sensors onto the lens with artificial intelligence (AI) technology that restores low-resolution sensor signals to high resolution.
The lens developed by the researchers integrates 100 light-detection sensors arranged in a 10x10 array. These sensors work by reading changes in light distribution as the eye moves, tracking the direction of gaze. The lens can distinguish not only up, down, left, and right directions but also diagonal movements, with this gaze information transmitted to the robotic arm to control its motion. Blinking enables the arm to grasp objects.
The research team also developed "meniscus pixel printing (MPP)" technology to directly print sensors onto the curved lens surface. The key to this technology is dabbing sensor material ink formed at the nozzle tip onto the lens surface. A meniscus refers to the convex or concave curved surface of a liquid, and this curvature creates a balance between the force expelling the ink and the force preventing ink spread, allowing precise amounts of ink to be deposited. When the ink dries, only the light-sensing perovskite material remains, functioning as the sensor.
Unlike conventional sensor fabrication, this new technology does not require masks for creating sensor patterns and can print sensors to match various eye curvatures, enabling individually customized lenses.
The problem of reduced signal resolution due to the limited space of a lens was solved through AI technology. Although only 100 physical sensors exist, deep learning-based super-resolution technology was applied to obtain signal data equivalent to having up to 6,400 sensors in an 80x80 array. The reconstruction time of just 0.03 seconds allows information to be transmitted to the robotic arm in near real-time.
In experiments using an eye model, the researchers successfully performed tasks including picking up and moving objects using only eye movements, achieving a directional recognition accuracy of 99.3%.
The research team explained that "this technology overcomes the spatial constraints of the ultra-compact form factor of a lens by combining hardware process innovation with AI-based signal restoration software technology." Professor Chung Im-doo stated, "We have demonstrated the feasibility of implementing an advanced human-robot interaction (HRI) system that directly converts human visual information into robot control signals without a separate controller," adding that "this has potential for expansion into various fields including augmented reality-based industrial robot remote control, exploration robot operation in disaster environments, unmanned systems and drone control in defense, medical and rehabilitation support systems, and smart mobility interfaces."
The research was conducted with Kong Byung-hoon and Kim Do-hyun, researchers in UNIST's Department of Mechanical Engineering, as co-first authors, and was supported by technology development programs from the National Research Foundation of Korea under the Ministry of Science and ICT, the Institute for Information and Communications Technology Planning and Evaluation, and the Ministry of Trade, Industry and Energy. The findings were published on March 11 in Advanced Functional Materials, a leading journal in materials science, and are scheduled to be featured as the front cover article of the upcoming issue.
