Inverse design of nonlinear response in flexible mechanical metamaterials

From wearable devices and energy-absorbing systems to footwear and soft robots, many applications would benefit from the inverse design of materials with a target non-linear mechanical response. Here, we present a framework to design flexible mechanical metamaterials with target nonlinear behavior. Our starting point is a metamaterial based on hinged rotating squares, which has recently attracted significant interest as it displays effective negative Poisson's ratio and supports the propagation of solitary pulses. We first show that a wide range of mechanical responses can be engineered by altering the shape of the quadrilateral units. Then, we use neural networks to learn the relationship between structure and mechanical response and combine them with an evolutionary approach to inverse design configurations with target non-linear stress-strain behaviors.