Researchers at The Hong Kong University of Science and Technology and partner institutions developed a hemispherical bionic eye that provides full-colour imaging and event-driven motion tracking, according to a study published in Nature Materials. The curved sensor integrates 367,500 total pixels at a pixel density of 1,905 pixels per inch.
Conventional planar vision sensors require multi-element optics and external processors to correct optical aberrations. That setup restricts miniaturisation, lowers power efficiency, and narrows the field of view. While curved sensors naturally mitigate aberrations in compact formats, previous devices suffered from low resolution that fell below practical operating levels.
Sensor capabilities
The device operates across a broad spectral range of 300 to 800 nanometres and delivers an aberration-corrected field of view exceeding 160 degrees. By utilizing a tandem artificial retina architecture, the sensor executes motion detection directly in-sensor rather than depending on external processing hardware.
Compared with standard frame-based visual recording, the event-driven system decreases bandwidth demand by over 99.95 per cent. During testing, the bionic eye achieved a motion recognition accuracy rate of 98.6 per cent.
Zhenghao Long and Zhiyong Fan conceived the system, which was supervised by Fan and Yunlong Zi across HKUST, HKUST (Guangzhou), and Fudan University. The authors stated that simulation and classification code is available on reasonable request, with underlying dataset files published alongside the paper.
