A hierarchical framework for classical and evolutionary robot control

Abstract

In nature, every living creature evolves. Humans are an example for the power of evolution. But evolution takes a long time for complex Problems, 200000 Generations of thousands of individuals, from the first ’homo’ to today. For complex tasks in evolutionary robotics this is much to long. In this diplom thesis I will use a hierarchical approach, breaking complex tasks down to planning and basic tasks. Then I will first evolve these basic tasks, making them available as parts for complex tasks.