While today technology that imitates human behavior is often used to increase a user's satisfaction and acceptance of robots, it hurts non-expert users in their ability to correctly anticipate a robot's capabilities, i.e. it produces a faulty mental model of the robot’s capabilities. In this project we develop novel ways of enabling users (1) to configure the robot behavior depending on their context and needs by (2) providing them with understandable and transparent information about the robot architecture by relying on observable and familiar concepts.