Biomimetic artificial neuromuscular fiber bundles with built-in adaptive feedback

Abstract

Skeletal muscles are composed of neuromuscular fiber bundles that combine the sensing capability of muscle spindle fibers with the actuation function of muscle fibers. However, it is difficult to develop artificial soft neuromuscular fiber bundles (NeuroMuscles) with sophisticated sensing-diagnosis-actuation autonomy. Herein, a unique rotational molding strategy is proposed to fabricate core-multishelled fibers with a liquid metal core, liquid crystal elastomer actuation layer, and adhesion sheath. The NeuroMuscles are developed by seamlessly welding multiple fibers through a self-reinforcing interface featuring independent channels for stimulus source and perception signals with built-in adaptive feedback. When integrated with NeuroMuscles, artificial arms and fingers can not only sense their own motion in real time but also detect the object’s surfaces. Importantly, the biomimetic knee-jerk reflex of artificial legs is achieved by establishing adaptive feedback within NeuroMuscles without off-board control systems for signal processing. The NeuroMuscles could function as indispensable components for implantable muscular reinforcements, next-generation soft machines, and beyond.