Abstract
Machines and robots play an ever increasing role in science and our society. However, so far they are laboriously planned and constructed for very specific tasks, usually at considerable expense. Current machines are specialised to only perform a very limited and fixed set of functions, which has greatly limited their autonomy. This brittleness contrasts starkly with the capabilities of animals that can adapt to a specific habit through adaptation by natural selection.
In this talk I will explain how simulations on HPCs, together with insights from robotics, artificial intelligence, and evolutionary biology, can help us to abstract and synthesize robotic artifacts that can adapt behaviorally and evolutionary to different environments. Together with the robot morphology we co-evolve advanced artificial neural network controllers enabled by GPU-accelerated Machine Learning techniques that can learn and reconfigure their structure mimicking plasticity mechanisms of animal nervous systems. This may facilitate automatic design of more intelligent machines and also allow us to answer open questions in evolutionary biology that are difficult to answer in vivo.