Improvement of assistive robot behavior by experience-based learning

Abstract

Robots designed for assisting humans in their homes need to adapt to the changing requirements of daily life. This requires multimodal sensor systems as well as learning strategies for understanding new goals and for recognizing new objects. However, coping with changes is not limited to environmental sensing. In order to achieve full autonomy, the robots must adapt their behavior due to good and bad experiences made. Concepts and first results of modelling intelligent sensing and adaptive behavior in an artificial mind as well as of merging mind and machine are presented in this paper.