Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
Evolutionary robotics is a promising approach to overcoming the limitations and biases of human designers in producing control strategies for autonomous robots. However, most work in evolutionary robotics remains solely concerned with optimizing control strategies for existing morphologies. By contrast, natural evolution, the only process that has produced intelligent agents to date, may modify both the control (brain) and morphology (body) of organisms. Therefore, co-evolving morphology along with control may provide a better path towards realizing intelligent robots. This paper presents a novel method for co-evolving morphology and control using CPPN-NEAT. This method is capable of dynamically adjusting the resolution at which components of the robot are created: a large number of small sized components may be present in some body locations while a smaller number of larger sized components is present in other locations. Advantages of this capability are demonstrated on a simple task, and implications for using this methodology to create more complex robots are discussed.