A Soft Shape-Morphing Surface with Embedded Shape Sensing Function

Soft robotic materials that can autonomously morph into desired shapes can enhance human-robot interaction. Recent advances in soft electronics and data-driven inverse design strategies have allowed for the creation of rapidly reprogrammable shape-morphing soft matter. To control these surfaces accurately and robustly, a shape-sensing capability is necessary. Current shape sensing methods heavily rely on exteroception, which can compromise the accuracy, reliability, flexibility, and speed of shape reconstruction in a deployed environment. In this work, we introduce a shape-morphing surface that can sense its own shape. The soft surface is constructed from silicone-based membranes containing interconnected stretchable silver nanowire conductors. The conducting meshes serve as both electromagnetic actuation elements in the presence of a magnetic field for shape-morphing and a resistive strain sensor for shape-sensing. We investigate the mechanics and materials of the soft composites and flexible sensors for optimized surface design for complex shape-morphing. Combining a mechanical model, finite-element analysis (FEA), and a machine learning pipeline with high-throughput experimentation enabled by the robotic surface, we develop a theoretical framework to describe the relationships between the input control voltages and output shapes, as well as the relationship between the deformed shapes and output sensing voltages. Stereo-imaging provides a ground-truth surface reconstruction and validates the accuracy of the embedded sensing with a maximum error of 5%. With the sensing and actuation functions closely integrated at the material level, the robotic surface is able to perform real-time closed-loop shape morphing in various changing environments.