Interactive Robotics in Medicine and Care
© Bielefeld University / Patrick Pollmeier
Reducing the Mental Model Mismatch in Cooperative Robot Teaching
Due to the increasing relevance of machine learning and artificial intelligence in everyday life, it is important to make the benefits accessible to inexperienced users. Of particular interest here are intelligent robots that are to be trained and taught by inexperienced users to perform specific tasks and support people in their activities. Intelligent assistance systems are also used in the fields of driving and medicine, where they are designed to adapt flexibly to the individual needs and wishes of users.
However, challenges exist in relation to the complex and difficult to understand nature of such systems. The vision of cooperative intelligence requires a mutual understanding between humans and AI systems to enable effective collaboration and achieve common goals. This project focuses on the role of mental models in human-robot interaction. Human mental models are a key component for understanding robots. The 'Theory of Mind' shows that efficient cooperation between humans and robots strongly benefits from a small discrepancy in mental models. Therefore, this project is concerned with better understanding people's perceptions of robots' capabilities. To this end, various feedback and transparency mechanisms are used to first quantify and then reduce the discrepancy in the human's mental model of the robot. This project is being realized in cooperation with the Honda Research Institute in order to contribute to the long-term improvement of human-robot interaction and to create a bridge between humans and machines.
