TOKYO, Nov. 28, 2025 — Researchers at Science Tokyo designed the world’s first automatic and adaptive, dual-mode LED-based optical wireless power transmission system, which operates seamlessly under both dark and bright lighting conditions. The system features AI-powered image recognition and can efficiently power multiple devices in order without interruption. Because the system is LED-based, it offers a low-cost and safe solution ideal for building a sustainable indoor Internet of Things infrastructure.
Optical wireless power transmission (OWPT) is a technology that reduces the need for traditional power delivery methods such as batteries and cable connections. In OPWT, energy is transmitted through free space, without physical wires, by converting electricity to light, transmitting it, and then reconverting light back into electrical power using photovoltaic receivers.
Past OWPT research has focused on laser-based systems; however, LED-based OWPT systems are more suitable for indoor IoT scenarios, as well as safer. However, these systems struggle with power loss over long distances and inconsistent performance under changing ambient lighting conditions. To combat power loss during long-distance transmission, the proposed system utilizes an adaptive lens system with a double-layer lens configuration, consisting of a liquid lens with a tunable focal length and an imaging lens. This setup automatically adjusts the beam spot size, based on the receiver distance and size, ensuring optimal power transmission.
To accurately aim at the light beam, the system uses an adjustable reflector that can be independently rotated in horizontal and vertical directions using two series-connected stepping motors. The researchers used a depth camera featuring an RGB sensor and an infrared (IR) sensor to ensure precise measurement of the photovoltaic (PV) receivers. The RGB sensor detects the PV receiver position, while the IR sensor identifies the beam’s irradiation spot. This allows the control system to adjust the reflector’s orientation towards the target receiver.
The PV receivers are equipped with retroreflective sheets around their edges. These sheets reflect IR light emitted by the depth camera’s IR projector, creating a clear outline for each PV receiver. This allows the system to accurately detect receiver shapes and positions, isolate the target PV area, and minimize interference from the surrounding objects. To further enhance accuracy, the researchers integrated a convolutional neural network based on the single-shot multibox detector algorithm.
The OWPT system can sequentially target multiple PV receivers of different sizes and at varying distances, rapidly switching between them without interruption. In experiments, the system seamlessly operated under both lit and unlit environments and achieved efficient, stable power transmission up to five meters.
The research was published in Optics Express (www.doi.org/10.1364/OE.574553).