Towards more efficient navigation for robots and humans

Effective robot navigation in the presence of humans is hard. Not only do human obstacles move, they react to the movements of the robot according to instinct and social rules. In order to efficiently navigate around each other, both the robot and the human must move in a way that takes the other into account. Failure to do so can lead to a lowering of the perceived quality of the interaction and, more importantly, it can also delay one or both parties, causing them to be less efficient in whatever task they are trying to achieve. In this paper, we present a system capable of creating more efficient corridor navigation behaviors by manipulating existing navigation algorithms and introducing social cues from the robot to the human. We give the results of a user study, demonstrating the effectiveness of our system, and discuss how it can be applied more generally to a wide variety of situations.