Japanese researchers have developed an indoor, LED-based wireless power system that uses adaptive optics and AI-guided beam control to power IoT devices up to 5 meters away, both day and night. The optical wireless power transmission (OWPT) system maintains a tightly focused LED beam for longer-distance energy transfer while precisely directing that light to multiple photovoltaic receivers, even during changing light conditions that can interrupt optical tracking and reduce power delivery.
Lead researcher Professor Tomoyuki Miyamoto from the Institute of Science Tokyo highlighted how LEDs offer a safer option for indoor wireless power transfer than the more widely developed laser-based systems. “The lower radiance and power density of LEDs make it easier to design systems that comply with optical safety requirements,” he told EE Times. “LEDs are also relatively cheap, have long operating lifetimes, and can be integrated into lighting-like modules.”
Miyamoto is also confident that his OWPT system will play a key role in safe and sustainable IoT infrastructure in buildings: “This is a stable and versatile wireless power transmission solution. Its most likely initial applications are indoor sensor networks in factories, farms, and smart buildings, where the delivered power can remain modest.”
Beyond RF limits
RF power transfer is already used in distributed IoT applications, including low-power sensor systems in buildings and warehouses, reducing reliance on batteries and wired connections. However, the technology faces issues around efficiency, range, and electromagnetic interference. For example, free-space propagation losses reduce the power available at the receiver as distance increases, while stray EMI emissions can disrupt connectivity.
View All In recent years, laser-based OWPT has emerged for longer-distance wireless power delivery, with systems demonstrating transmission over tens of meters. By concentrating energy into a narrow optical beam, laser approaches can achieve higher power densities than RF systems while avoiding RF interference concerns. However, safety considerations remain a major barrier for indoor deployment, creating interest in LED-based OWPT.
Miyamoto and his team unveiled their first LED-based approach back in 2019, which used collimating and focusing optics to efficiently transmit optical power across 1 meter. But as Miyamoto pointed out, longer transmissions suffered severe power attenuation as the LED beam diverged, reducing optical power density at the receiver, while the system also struggled to operate reliably under varying lighting conditions. The latest AI-guided adaptive optics system addresses these limitations.
In a bid to overcome power attenuation problems and extend transmission distances beyond a meter, the researchers used an adaptive double-layer lens, comprising a liquid lens with a tunable focal length and a fixed imaging lens. These more sophisticated beam-shaping optics were designed using ray-tracing simulations to balance key parameters such as LED beam collection, the actual beam spot size, and transmission distance.
“By dynamically adjusting the liquid-lens focal length to suppress [LED] beam expansion … effective power transmission up to 5 meters has been demonstrated,” Miyamoto said. “Also, the optical power received at 3 meters was around 8× higher than the previous fixed-optics system.” Miyamoto also pointed out how simulations indicated that the LED beam could remain focused for up to 10 meters, although he and colleagues haven’t yet demonstrated this in their experiments.
To direct the tightly focused LED beam toward photovoltaic receivers, the system uses a motorized reflector that is adjusted according to data captured by an RGB-IR depth camera. The camera identifies the receivers and determines where the beam should be directed based on their size and location. On-board AI, using a convolutional neural network based on a single-shot object-detection algorithm, further hones detection accuracy.
To ensure operation in darkness, retroreflective sheets—commonly used on traffic signs and safety clothing—are attached to the edges of the PV receivers to reflect light from the depth camera’s IR sensor and create a clear outline of each receiver. “The reflected pattern is captured with high contrast, allowing the system to locate a PV receiver even when the room lighting is switched off,” Miyamoto said.
Five-meter demo
Experimental tests look good. Miyamoto and his team placed five different-sized PV receivers, 2 to 4 meters apart, around their lab, powering each one sequentially. Their LED-based OWPT system was able to locate and seamlessly switch between each receiver in both light and dark conditions, efficiently transmitting power up to 5 meters.
“Under illuminated conditions, medium and large receivers were detected at up to 5 meters, while under unlit conditions, retroreflector-equipped receivers were reliably detected at approximately 4 meters,” Miyamoto said.
However, as he also noted: “Our LED system is primarily intended for meter-scale indoor applications. For transmissions over tens of meters or when high power is needed, a laser-based system becomes more practical.”
Further improvements are also possible. According to Miyamoto, the overall optical system efficiency came in at just over 56%, but the researchers now intend to raise this to more than 80%. Possible routes forward include using a more focused LED or mounting a microlens onto the LED to reduce the spread of light and get more light into the system’s beam-shaping optics. The researchers also hope to characterize the relationship between the temperature of the liquid lens and its focal length so they can introduce thermal compensation to prevent the focal drift that reduces receiver tracking precision.
“As well as lens-system efficiency, our next steps are to increase recognition robustness, integrate the optical source and power electronics, and evaluate long-term reliability,” Miyamoto said. “If we can now get an industry partner to look at product integration and qualification, we will be able to build a pilot system relatively quickly.”
Read also:
[Monitoring the Output Current of Offboard LED Drivers](https://www.powerelectronicsnews.com/monitoring-the-output-current-of-offboard-led-drivers/)
[Inside CEA-Leti’s Push to Industrialize MicroLED Interconnects](https://www.eetimes.com/inside-cea-letis-push-to-industrialize-microled-interconnects/)
[The Security AI That Learns the Language of Movement](https://www.eetimes.com/the-security-ai-that-learns-the-language-of-movement/)
[How Much Impact Will AI Have on IoT Software Engineering?](https://www.eetimes.com/how-much-impact-will-ai-have-on-iot-software-engineering/)