Sunny Optical Expands Optics Plus AI Push Into Robots and Smart Glasses

The push comes as two device categories approach mass production. Counterpoint data cited in the report showed global humanoid robot shipments exceeded 22,000 units in the first half of this year, up nearly 300% from a year earlier. IDC data showed global smart glasses shipments reached 3.566 million units in the first quarter, up 130.1% year on year. Although the products are different, both require optical systems that see more clearly and weigh less, the report said.

Humanoid robots need to sense obstacles and distance to plan actions, AI camera glasses must restore a first-person view within thin frames, and AI plus AR glasses need to overlay digital information onto the real world. Lenses, camera modules and waveguides therefore determine perception accuracy, display quality and device form.

Xiong Guanhao, senior director of strategy and marketing at Sunny Optical, said the value share of vision systems in intelligent hardware is rising. He pointed to lawn mower robots as an example. In the past, some automatic mowers relied on GPS differential positioning to follow planned routes. Positioning could tell a machine where it was, but it struggled to identify surrounding obstacles and grass areas, and route planning could miss sections, according to the report. Sunny Optical provides an X-BOX mower vision solution with 360-degree panoramic cameras, active binocular cameras and RGB cameras. While cameras collect the surrounding environment, perception algorithms identify obstacles and grass areas, and path planning then decides the next move.

Vision sensing cannot solve all scenarios at once. Robot 3D vision faces failures in ranging, interference, depth distortion and complex calibration, the report said. Outdoors, strong light, high reflection, low reflection, small objects and translucent materials further complicate stable depth and spatial information. Humanoid robots face sharper problems. In the several tens of thousands of humanoid robots shipped in the first half, intelligent manufacturing accounted for 13% and warehousing and logistics 5%. As real work scenarios increase, whether capabilities trained in fixed environments can be reused becomes the next hurdle. The report cited Unitree founder Wang Xingxing as saying the industry's critical point has two 80% thresholds: if a robot can complete about 80% of tasks in about 80% of unfamiliar scenarios through language instructions alone, it will approach embodied AI's ChatGPT moment.

Sunny Optical is responding with more than a single camera. It has laid out 3D vision routes including binocular, iToF, dToF and structured light for different scenarios, and added multi-sensor fusion, calibration and algorithm capabilities. Xiong said many emerging industry customers may not be good at optical hardware tuning or optical algorithm matching and need a complete vision system that can be directly integrated and deployed. Sunny Optical's 2026 interim results showed its mower robot and warehouse automation robot projects have entered volume production and it has won several new complete machine and system solution projects. The company, which says it has worked in robot vision for more than 11 years, has sent core vision hardware such as ToF cameras and binocular vision to leading robot makers for sample testing.

Smart glasses require optics not only to see clearly but to look natural. A smart glasses CEO said in a public interview that optical display remains a key technical direction, with waveguide glasses still facing issues such as light transmittance, rainbow patterns and light leakage. Another AI glasses founder said one difficulty in reducing weight below 50 grams lies in the evolution of underlying optical technology. Choices about color display methods, monocular or binocular design and camera count also affect total weight, according to the report.

Sunny Optical has focused on smaller optical engines, thinner lenses and more highly integrated modules. It recently disclosed high-performance lightweight prisms, ultra-thin VID lenses, an extremely miniaturized display optical engine and high-precision MicroLED packaging. Its VID lens has reached a mass-producible thickness of less than 0.8 mm, providing support for reducing device weight. For AI camera glasses without displays, Sunny Optical previously disclosed a partnership with Meizu's first camera-focused AI smart glasses, StarV Snap, to provide camera modules and a one-stop smart glasses system solution. In AR products with displays, its thin and light full-color AR display module entered small-batch production in the first half of 2026.

Xiong does not believe every system solution is worth pursuing. He repeatedly cited two thresholds: whether core optical components account for a high enough share of the product's value and whether the segment is something Sunny Optical is truly good at.

As a full-scenario, end-to-end intelligent optical system provider, Sunny Optical continues to upgrade precision manufacturing while extending toward systems. Xiong said system solutions and large-scale manufacturing are capabilities that must advance together. The company entered mobile phone camera module manufacturing in 2003 and later expanded into actuators and precision assembly and alignment. In the AI era, that experience has given it transferable capabilities. Its AA active alignment process from the phone era is now used in precision production of robot vision modules across scenarios. Ultra-thin packaging experience from phone camera modules supports the miniaturization of AR optical engines; Sunny Optical applied such packaging technology to AR optical engine development and designed what the report described as the industry's smallest AR optical engine solution.

Manufacturing scale matters as well. Shipments of millions or tens of millions of units in the phone industry helped Sunny Optical build a mature large-scale delivery system. When robot customers need not just a lens but an entire vision system, the company can respond quickly with modules plus SDKs, or even complete devices, reducing integration difficulty for customers. Xiong said Sunny Optical is also using AI models in production to improve yield and efficiency and to make line adjustments and responses to customer demand faster. The company's goal is not only to sell more optical components into new AI-era terminals, but to participate earlier in product and system definition and place optical capabilities closer to the core of next-generation intelligent hardware and AI infrastructure.