Socially-Aware Navigation Using Topological Maps and Social Norm Learning

We present socially-aware navigation for an intelligent robot wheelchair in an environment with many pedestrians. The robot learns social norms by observing the behaviors of human pedestrians, interpreting detected biases as social norms, and incorporating those norms into its motion planning. We compare our socially-aware motion planner with a baseline motion planner that produces safe, collision-free motion. The ability of our robot to learn generalizable social norms depends on our use of a topological map abstraction, so that a practical number of observations can allow learning of a social norm applicable in a wide variety of circumstances. We show that the robot can detect biases in observed human behavior that support learning the social norm of driving on the right. Furthermore, we show that when the robot follows these social norms, its behavior influences the behavior of pedestrians around it, increasing their adherence to the same norms. We conjecture that the legibility of the robot’s normative behavior improves human pedestrians’ ability to predict the robot’s future behavior, making them more likely to fol-