Abstract

Mobile robot navigation has seen extensive research in the last decades. The\naspect of collaboration with robots and humans sharing workspaces will become\nincreasingly important in the future. Therefore, the next generation of mobile\nrobots needs to be socially-compliant to be accepted by their human\ncollaborators. However, a formal definition of compliance is not\nstraightforward. On the other hand, empowerment has been used by artificial\nagents to learn complicated and generalized actions and also has been shown to\nbe a good model for biological behaviors. In this paper, we go beyond the\napproach of classical \\acf{RL} and provide our agent with intrinsic motivation\nusing empowerment. In contrast to self-empowerment, a robot employing our\napproach strives for the empowerment of people in its environment, so they are\nnot disturbed by the robot's presence and motion. In our experiments, we show\nthat our approach has a positive influence on humans, as it minimizes its\ndistance to humans and thus decreases human travel time while moving\nefficiently towards its own goal. An interactive user-study shows that our\nmethod is considered more social than other state-of-the-art approaches by the\nparticipants.\n