Evolutionary Morphology for Real Cubic Modular Robots

Recently modular robots have become more capable of practical applications with the recent improvement of sensors and actuators. Among them, a robot has been developed which can change its pattern according to the landscape or for its own purpose. However, deciding the appropriate pattern and controller manually is not easy with the increased number of modules. In this paper, we propose a new approach to automatically building patterns of block-type robots by means of artificial-life morphogenesis. We empirically show the emergence of the effective patterns in both virtual and real worlds, some of which seem to be surprisingly counter-intuitive.