Neuromorphic Metamaterials for Mechanosensing and Perceptual Associative Learning

Physical systems exhibiting neuromechanical functions promise to enable structures with directly encoded autonomy and intelligence. A neuromorphic metamaterials class embodying bioinspired mechanosensing, memory, and learning functionalities obtained by leveraging mechanical instabilities integrated with memristive materials is reported. The prototype system comprises a multistable metamaterial whose bistable dome‐shaped units collectively filter, amplify, and transduce external mechanical inputs over large areas into simple electrical signals using embedded piezoresistive sensors. Dome deformations in nonvolatile memristors triggered by the transduced signals, providing a means to store loading events in measurable material states are recorded. Sequentially applied mechanical inputs result in accumulated memristance changes that allow us to physically encode a Hopfield network into the neuromorphic metamaterials. This physical network learns the history of spatially distributed input patterns. Crucially, the neuromorphic metamaterials can retrieve the learned patterns from the memristors’ final accumulated state. Therefore, the system exhibits the ability to learn without supervised training and retain spatially distributed inputs with minimal external overhead. The system's embodied mechanosensing, memory, and learning capabilities establish an avenue for synthetic neuromorphic metamaterials that learn via tactile interactions. This capability suggests new types of large‐area smart surfaces for robotics, autonomous systems, wearables, and morphing structures subjected to spatiotemporal mechanical loading.