Seminář UPSY - Trefzer M.A.: From Biology to Hardware - Can Social Insects Organise Distributed Systems?
Abstract: The increasing versatility, performance, compactness and power efficiency of today's electronic systems is pushing technology to its physical limits, making designing robust systems extremely challenging. Biological organisms have long since accomplished the feat of operating reliably with highly variable components, as well as maintaining and tuning themselves in changing environments, when faults occur or they are otherwise perturbed. Such biological mechanisms inspire how hardware could evolve and how electronic systems could self-organise and self-repair. Evolutionary systems are about hardware and software that can autonomously adapt their structure and behaviour in order to optimally carry out specific tasks under changing conditions, taking inspiration from biological organisms with evolution as nature's guiding principle. Circuits can be evolved from in silico Primordial Soup, shape evolves into function and unexpected material properties are uncovered and made useful. In nature there are many examples of systems that, unlike traditional computing architectures, cope well with having thousands of computing elements, social insects being one of them. Starting with a single individual, a queen, many social insect colonies quickly grow to hundreds of thousands of cooperating individuals with a moderate amount of intelligence and without central control. In this talk I will introduce many-core computing platforms, Centurion and Graceful, and discuss how algorithms inspired by such insects could be applied to produce self-organising, self-optimising and self-healing "colonies" of computing platforms in the future, comprising of hundreds of nodes.
Bio: Martin is a Senior Lecturer (Associate Professor) in the Department of Electronics at York. His research interests include variability-aware analogue and digital hardware design, biologically motivated models of hardware design, evolutionary computation, and autonomous fault-tolerance. His vision is to create novel architectures and autonomous systems, which are dynamically self-optimising and inherently fault-tolerant, by porting key enabling features and mechanisms from nature to hardware. He is co-investigator on 3 currently running EPSRC / DSTL projects: Platform Grant - Bio-inspired Adaptive Architectures and Systems (EP/K040820/1), Graceful (EP/L000563/1) and Complex In-materio Computation for Robust Dynamical Control, as well as on the previous EPSRC project PAnDA (EP/I005838/1). He is a senior member of the IEEE, a member of the DPG, co-chair of the International Conference of Evolvable Systems (ICES), and vice chair of the IEEE Task Force on Evolvable Hardware.