The goal of the position is to create prediction methods for proactive planning of future robot actions and to design robot acting mechanisms for adaptive response ranging from quick and intuitive to slower well-reasoned. We combine sensing across multiple modalities with learned knowledge to predict outcomes and choose the best actions. The goal is to transfer these skills to human-robot interaction in home scenarios, including the support of everyday tasks and physical rehabilitation. Thus, it is relevant to work with implementation and research within robot perception and control for the robot tasks. User studies through human-robot interaction experiments are to be performed. A PhD fellow and a researcher are already hired for the project and will complement in performing the above outlined research.