Enabling long-term human-robot interaction through adaptive behavior coordination

Robots in domestic environments currently perform simple tasks such as floor cleaning in a fully autonomous fashion with great success. During those tasks the lack of human-robot interaction (HRI) is given by design, but is also limiting the robot's activities when an unforeseen situation occurs. Based on state-of-the art research in attention methods and sliding autonomy of social robots, we develop a behavior modeling framework, for the recognition of situations that the robot can not mitigate autonomously and the introduction of the user as a partner in the chosen mitigation strategy. We have already explored support for elderly in their homes over the course of three weeks with a fully autonomous robot, during which we discovered the need for adaptive autonomy and behavior. The developed framework will be evaluated in a further study, to show if it can be used across different robots without being penalized by the users after the robot needed assistance, which we see as a must to overcome the novelty effect to further enable long-term HRI in domestic environments.