Evolvable hardware for fault-tolerant applications

Evolution is perceived by the change of state in a dynamic complex system. In this broader sense, we can apply the attribute “evolvable” to numerous man-made systems whose configuration or response is influenced by the environment. Genetic algorithms, neural networks, artificial intelligence projects and embryonics can all be classified as evolvable systems. By adopting the mechanisms underlying the developmental process of embryos, embryonic systems present interesting characteristics that make them an attractive alternative for the design of fault-tolerant systems. The embryonics approach is coherent with a recent rise in the interest shown by the scientific community for applying biological concepts to the solution of engineering problems. We demonstrate with an example that embryonic systems posses the attribute of fault tolerance in the same way as biological systems do. In multicellular organisms, the failure (death) of one cell does not impact on the overall functionality of the tissue/organ/limb affected because of the tremendous amount of redundant elements available. If a cell or small group of cells die, their function is replaced by their healthy neighbours. Embryonic arrays seem to successfully mimic this mechanism. We still are in the early stages of the embryonics project, but the results obtained so far are encouraging to keep investigating on the application of biological concepts to the design of engineering systems