A Biomimetic Self-Adaptive Neurovision Eye With an Integrated Gel Iris and Retinamorphic Architecture.

Biological vision systems integrate light perception with self-adaptive regulation to ensure a stable view. Mimicking this in an artificial neuromorphic system requires synergistic components for dynamic light control and in-sensing processing. Here we present a biomimetic self-adaptive neurovision system featuring an integrated gel iris and retinamorphic architecture. The hydrogel iris, structured as a double concentric ring, autonomously modulates its opacity from 2.95 to 89.85% via a photothermally driven themochromic effect, while the ionogel retina, designed as a hemispherical micropillar array with a photosynaptic ionotronic heterostructure, enables broadband photoperception (360-680 nm), visual learning (paired-pulse facilitation, 118%), and self-powered operation. This synergistic integration allows the eye system to maintain high-fidelity image recognition (accuracy, 97.2%) under strong light exposure, paving the way for emerging zero-powered soft visual prosthetics and machine vision.