Mercoledì 3 luglio alle ore 11:30 presso l’aula Grandori (Edificio 4, piano seminterrato, piazza Leonardo da Vinci 32) si terrà un seminario dal titolo: “Cognitive dynamical systems: Talking structures, smart nano-robots and the limits of green computers“, tenuto da Marc Serra-Garcia (tenure-track group leader at the AMOLF institute on the physics of complex matter in Amsterdam, the Netherlands).

Abstract

The dynamic response of elastic structures can be very complex; surprising us with instabilities, parametric resonances and chaotic dynamics. This complexity is responsible for unpredictable failure mechanisms and numerical modelling challenges. But could we use the richness of behaviour to our advantage? Can we design structures that solve problems using solid mechanics instead of electronics and software? In this talk, I will show two examples of our recent work in cognitive elastic structures. First, I will present a passive device that recognises spoken commands, built by implementing machine-learning models in networks of elastic resonators. Such structure can detect relevant events (e.g. someone saying ‘OK Google’) while consuming zero standby power, opening the door to smart devices (pacemakers, sensors, smart speakers) that do not require batteries to operate. Second, I will show a mass-spring system that executes an algorithm to find the exit of a maze, generated automatically from from a text description of the desired behaviour — written in a ‘mechanical programming language’. Such structure could one day form the basis for smart robots with no control electronics. A strong point of intelligent elastic structures is their high energy efficiency. At the microscopic scale, thermodynamics restricts the computations that we can we perform, given a fixed amount of input energy. To conclude the talk, I will briefly show our recent simulations on the scaling-down of elastic intelligent structures. Our result suggests that, while regular electronic computers consume around four orders of magnitude more power than fundamentally necessary, the thermodynamically-limited regime may be accessible to smart elastic structures within the capabilities of current fabrication technology. While mechanical forms of information technology will most likely not replace electronics in general purpose computers, the ability to function without a battery, transducers or toxic materials makes cognitive mechanics a promising solution for a remarkable number of problems and use cases — such as implantable sensors, zero-power IoT devices and other power or size constrained applications.

Speaker’s bio

Marc Serra-Garcia is a tenure-track group leader at the AMOLF institute on the physics of complex matter in Amsterdam, the Netherlands — A position he took after his studies at Caltech and ETH Zurich. His research focuses on developing elastic structures with novel functionalities by combining advanced design algorithms and precision microfabrication with insights from fundamental physics. His work includes the invention of a nonlinear system that extracts energy from random vibrations, the demonstration of wave-controlling materials based on topological insulators, and, currently, the development of elastic computing structures — from speech classifiers to entirely-mechanical microprocessors; area of research that has recently been recognised by an ERC Starting grant.