A hypersensitive neuromorphic airflow sensor inspired by vision-compensatory scorpion mechanoreceptors for respiratory pattern analysis.

Abstract

Efficient acquisition of spatial airflow information is vital for organisms to orient within complex environments and detect predators. For scorpions with degraded vision, specialized mechanosensory trichobothria provide a crucial vision-compensatory mechanism, enabling hypersensitive perception of subtle airflow fluctuations. Inspired by this evolutionary adaptation, we present a biomimetic neuromorphic airflow sensor (BNAS) integrating a bioinspired lever-amplification structure with a pressure-induced ionic enrichment mechanism. This synergistic design inherits the hypersensitive anemosensation and neural response features of scorpion. The BNAS demonstrates a superior sensitivity of 18.22% (m/s)^-1 at low velocities and maintains high performance across a broad dynamic range (0.1 to 10.27 m/s), along with omnidirectional detection capability. The integration of this neuromorphic hardware with AlexNet deep-learning algorithm enables the efficient extraction of human respiratory patterns, achieving 95.56% accuracy in identifying individual "breathing fingerprints." Our work underscores the potential of bioinspired neuromorphic systems to bridge the gap between biological perception and artificial sensing, establishing a neuromorphic front-end design paradigm that advances next-generation brain-inspired computing.