Abstract

Most work in evolutionary robotics centres on evolving a controller for a fixed body-plan. However, previous studiessuggest that simultaneously evolving both controller and body-plan could open up many interesting possibilities. However, thejoint optimisation of body-plan and control via evolutionaryprocesses can be challenging in rich morphological spaces. Thisis because offspring can have body-plans that are very differentfrom either of their parents, leading to a potential mismatchbetween the structure of an inherited neural controller and thenew body. To address this, we propose a framework that combinesan evolutionary algorithm to generate body-plans and a learning algorithm to optimise the parameters of a neural controller. The topology of this controller is created once the body-plan of each offspring has been generated. The key novelty of the approach is to add an external archive for storing learned controllers that map to explicit ‘types’ of robots (where this is defined with respect to the features of the body-plan). By initiating learning froma controller with an appropriate structure inherited from thearchive, rather than from a randomly initialised one, we show that both the speed and magnitude of learning increases over time when compared to an approach that starts from scratch, using two tasks and three environments. The framework also provides new insights into the complex interactions between evolution and learning