Continuous time recurrent neural networks: a paradigm for evolvable analog controller circuits

Abstract

This paper argues that Continuous Time Recurrent Neural Networks (CTRNNs) provide a particularly attractive paradigm under which to evolve analog electrical circuits for use as device controllers. It will make these arguments both by appeal to existing literature and by the example of a successful project in the control of an autonomous robot. The paper will conclude with a discussion of future work and goals.