A Foundation for Robot Learning

Abstract

This paper considers the fundamental issues of robot learning in which answers to basic questions on robot learning, such as "What can the robot learn?", "What are the consequences of robot learning?", "How does the robot learn?", "How fast do robots need to learn?", and "When do robots learn?" are addressed. The answers to these questions may lead to the identification of the elements of robot learning and the interaction between these elements. Hence, the purpose of this paper is to discuss the fundamental issues in a holistic manner so that key elements that characterise robot learning can be formalised into a framework.