Title:

Distributed Intelligence and Control in Adaptive Kirigami Microrobots

Abstract:

Imagine a nano-meter thin sheet intelligently programming itself to swim and walk by folding into complex 3-dimensional shapes. While this sounds like a dream for most microrobots, a pioneering group at Cornell is trying to bring this dream to life by embedding a distributed array of integrated circuits (ICs) across a nano-meter thin actuating kirigami sheet. This novel class of intelligent elastronic metamaterials computes through elastic shape changes processed by simple electronic circuits, which raises two fundamental theoretical questions at the intersection of computation and mechanics. First, how do we encode a range of complex behaviors using a distributed network of integrated circuits with extremely limited computing power? Second, how do we integrate dynamical computation with mechanical shape change? I will provide possible answers to these two questions by presenting on my Ph.D work. In one project I developed new theory for embedding a complex and controllable range of dynamical trajectories in recurrent neural networks (RNNs), thereby allowing me to treat each integrated circuit as a simple electronic neuron in a distributed kirigami RNN. I will show how a similar approach may allow for answering the first question by creating simple encoding of complex dynamics. I will also present on a new theory I developed for designing the geometry of flexible degrees-of-freedom in metamaterials, thereby allowing me to define the configuration space of kirigami actuations and hence the allowed computations. As a postdoc, I propose to combine my prior work in RNNs and metamaterials, to develop a new theory for distributed intelligence and control in adaptive kirigami microrobots.