Soft Robotic Skin from Intelligent Meta-Materials

Smart product design has traditionally focused on design paradigms that facilitate the design of mechanisms with attached sensing, attached computation and attached actuation. The field of meta-materials, on the other hand, investigates imbuing materials with novel characteristics, such as sensing, intrinsically to the material. Structured Computational Polymers (SCP) are proposed intelligent meta-materials that combine sensing, cognition, and actuation into bulk materials, bridging the gap between novel material characteristics and integrated design paradigms. This paper describes the development of a flexible polymer meta-material that embeds a neuromorphic architecture for computation based on printable organic semiconductors with intrinsic pressure sensing co-designed with an algorithm to compute the centroid of pressure. Based on a synthetic neural network with quantized weights, the neuromorphic architecture is trained to achieve the desired computation in simulation and optimized offline to account for the practical constraints of the polymer architecture. The polymer transistors and memristive storage elements of the artificial neural network are emulated in this prototype skin with conventional silicon circuits as an intermediate proof-of-concept as we transition from lithography to polymer printing to produce a skin that exhibits the desired characteristics, enabled by our co-design framework. A limited fraction of a skin sheet.