A behavior-based architecture for teaching more than reactive behaviors to mobile robots

The article gives an overview of learning aspects regarding behavior based control architectures for autonomous mobile robots. We propose a common, modularized architecture, which is particularly suited for learning experiments. The main focus lies in "Learning from Demonstration" in a spatial domain, which means teaching motor behaviors by humans or other robots. First results applying RBF approximation, growing neural cell structures and probabilistic models for progress estimation are presented.