EPREVAEngineer Paper
SENSORSInventionElectronicsMaterialsRobotics & automation

HKUST builds hemispherical bionic eye with 367,500 pixels

The curved retina combines full-colour imaging over a field of view exceeding 160 degrees with in-sensor motion detection that cut bandwidth demand by more than 99.95 % in the study's tests.

hemispherical curved sensor chip, dense pixel array on the curved surface, gloved hand or lab probe holding it, cleanroom or optics lab setting
AI-generated image

What was built

Researchers at The Hong Kong University of Science and Technology (HKUST) have built a hemispherical tandem artificial retina with 367,500 total pixels and a pixel density of 1,905 pixels per inch (ppi), according to the university. The device combines full-colour imaging, an aberration-corrected field of view exceeding 160 degrees and event-driven motion detection.

Professor Fan Zhiyong led the work. He is chair professor of the Department of Electronic and Computer Engineering and the Department of Chemical and Biological Engineering, and director of the State Key Laboratory of Displays and Opto-Electronics at HKUST.

According to the university, the system offers a pathway toward compact vision sensors for future robots and autonomous systems, enabling machines to capture environmental information while reducing the data required for motion recognition. The findings were published in Nature Materials in a paper titled 'An aberration-corrected bionic eye for full-color vision and event-driven motion detection'.

How it works

Conventional lithographic processes are primarily designed for flat substrates, so the HKUST team developed a 3D laser fabrication approach that enables dense arrays of light-sensitive pixels to be fabricated on a hemispherical substrate, according to the university.

The tandem artificial retina incorporates blue-, green- and red-sensitive layers. It provides full-colour imaging with an overall spectral response spanning 300 to 800 nanometres, and it maintains aberration-corrected imaging across a field of view exceeding 160 degrees.

The tandem design also enables in-sensor, event-driven motion detection. Conventional frame-based approaches obtain motion data by processing sequences of complete images. The new system detects motion-related information within the sensor, eliminating the need to transmit full image sequences for the tasks evaluated.

How it was tested

Under the study's test conditions, the university reports, the event-driven approach reduced bandwidth demand by more than 99.95 % compared with frame-based imaging, while achieving a motion recognition accuracy of 98.6 %.

The pixel count is a step up from the team's earlier device. Fan said the first-generation bionic eye developed in 2020 contained only 100 sensing pixels, against 367,500 in the current system. The university characterises the increase as an important step toward the practical application of hemispherical vision sensors.

What is not yet proven

Fan said that with continued advancements the team believes it can develop a next-generation bionic eye with over 1 million sensing pixels, opening new opportunities for practical applications in robotics and autonomous driving.

Build details

The paper is published in Nature Materials (DOI: 10.1038/s41563-026-02685-5). The first author is Long Zhenghao, a former doctoral student and postdoctoral researcher in HKUST's Department of Electronic and Computer Engineering, now an assistant professor at Fudan University. Fan served as the sole corresponding author, and the research was conducted in collaboration with Professor Zi Yunlong of HKUST (Guangzhou).

Build sheet

  • 367,500 total pixels in the hemispherical artificial retina
  • 1,905 pixels per inch (ppi) pixel density
  • Field of view exceeding 160 degrees, aberration-corrected
  • Spectral response spanning 300 to 800 nanometres (blue-, green- and red-sensitive layers)
  • More than 99.95 % reduction in bandwidth demand versus frame-based imaging, under the study's test conditions
  • 98.6 % motion recognition accuracy, under the study's test conditions
  • 100 sensing pixels in the team's 2020 first-generation device

Sources

  1. An aberration-corrected bionic eye for full-colour vision and event-driven motion detection — Nature Materials — Zhenghao Long, Kaichen Wang, Shaoshuai He et al.
  2. High-resolution bionic eye combines color imaging with more efficient motion detection — TechXplore (Engineering)

Discussion

Technical questions and corrections are welcome. Comments are signed and checked before they appear.

No comments yet.

Sign in to add a typed review comment. Reviews are signed with your name.

More inventions