A dynamically reprogrammable metasurface with self-evolving shape morphing

Achieving intelligent soft matter that can self-configure, self-adapt, and even self-evolve is of great importance for many engineering sciences. The past decade has witnessed phenomenal investment in developing responsive metamaterials that can swiftly shift their structures, and henceforth their performing functions. However, creating materials and structures with precise, complex, fast, and reversible programmability and incorporating real-time feedback remains challenging. In this work, we describe a dynamically reprogrammable mechanical metasurface with embedded actuation, sensing, and feedback-control functions for shape-morphing intelligence. The voltage-controlled Lorentz force driving a flexible, conductive 2D mesh in a static magnetic field lays the foundation for a highly integrable and scalable digital-physical interface. A custom-built stereo-imaging setup incorporating multiple webcams enables an in-situ measurement of the 3D out-of-plane deformation of the 2D precursor. With an implementation of a programmable control guided by an optimization algorithm, the metasurface acquires the ability to self-evolve to approach a given target shape in absence of presuming models. The closed-loop structure opens opportunities for physical simulations on non-linear systems. Such an experiment-driven method also exhibits superior advantages in morphing against extrinsic or intrinsic perturbations that theoretical models cannot predict or account for, including environmental changes, external loading, and sample defects. In addition to the morphing intelligence, the metasurface obtains a functional intelligence to adapt its morphology to accomplish prescribed tasks, bypassing any involvement of shape designs. Upon assignment of aims, the direct goal-oriented morphing also allows the metastructure to attain multifunctionality with an ability to decouple naturally coupled structural functions.