Stretchable Hydrogel Optical Memristor for Photonic Near‐Sensor Neuromorphic Skin

Abstract The development of skin‐like optical memristors is crucial for wearable photonic neuromorphic computing, which enables ultrafast artificial neural networks for edge information processing. However, most existing optical memristors are made of rigid materials, posing challenges in integrating with soft photonic sensors and increasing the risk of device failure under large strains. Additionally, the strain‐perturbed photonic matrix‐vector multiplier (PMVM) may malfunction when stretched. In this paper, a design strategy that utilizes transverse surface colorization (TSC) of hydrogel waveguides is presented to create stretchable optical memristors that can implement a strain‐invariant PMVM. These stretchable optical memristors provide more than 4‐bit distinct, non‐volatile memory states (>10 4 s). By utilizing stretchable optical memristors as bending sensors and strain‐invariant PMVM, an integrated photonic near‐sensor neuromorphic skin capable of real‐time hand gesture recognition (latency < 0.5 ns) is demonstrated with 95.5% accuracy even under 30% strain. This seamless integration of sensing, memory, and processing capabilities in a photonic near‐sensor neuromorphic skin paves the way for future wearable photonics.