High-level synthesis of soft robotic structures

To build soft robots that can process information and make autonomous decisions, we must encode intelligent responses directly in the structural dynamics, taking advantage of multistability and snapping to store state information and implement logic functions. However, elastic structures are described by partial differential equations that are both nonlinear and dissipative. Therefore, the inverse problem of designing a system for a particular response is extremely hard. In this work, we design elastic structures that execute arbitrary algorithms. We do so by taking the notion of hardware-description-language (HDL) from electronics: We describe the algorithm that the robot will execute using a mechanical description language (MDL), that we then convert to a geometry in a multi-step mechanism (Code > Logic gates > Nonlinear mass-spring model > Geometry). We demonstrate this by designing an elastic system that implements a graph search algorithm to drive a soft robot out of a maze.