If robots are to be able to assist humans with a wide range of complex tasks in dynamic real-world environments, such as within their own homes, these robots must continuously learn from their respective users. An interesting question in this context is which interaction capabilities need to be pre-programmed and which should be acquired through learning based on explanations or experience. A human-centered approach first requires an understanding of how people interact with, perceive, and guide robots over an extended period of time. From this, ideas can be developed on how to enable robots to provide goal-oriented assistance.

That is why I am interested in the practical use of robots in social contexts within real-world environments involving real people. The opportunity to learn by observing the reactions of people who have little or no experience with robots is extremely valuable to my work. Their feedback, along with the robots’ inevitable mistakes, helps me improve the applications I develop. Such a reality check directly highlights existing problems and is indispensable for making improvements.