³ Synthesis of smart structures and soft robots

Soft robots use structural dynamics to sense the environment, process information and respond to stimuli; nonlinear elasticity allows them to complete tasks without control electronics. Unfortunately, it is very difficult to design robotic structures with the sufficiently complex dynamics required to accomplish nontrivial goals, such as navigating through unknown environments. We aim to solve this problem by inventing novel computational methodologies that generate smart robotic structures algorithmically

Our approach starts by inventing mechanical programming languages that describe the algorithm that the structure must execute. We then build compilers that can translate our mechanical source code into structural geometries. After setting up this workflow, we can design structures that perform very complex tasks. For example, we can implement a Depth-First-Search algorithm in a mechanical programming language, and compile it into a robot that can find the exit of a maze. This approach draws inspiration from the design of very-large scale integrated circuits (VLSI). In VLSI design , a circuit description, written in a hardware description language, is translated to a circuit consisting of millions or billions of transistors — a task that would be entirely impossible without modern automated tools.