Time dependent planning on a layered social cost map for human-aware robot navigation

As robots make their way into our everyday lives, new behavioral concepts are needed to assure their acceptance as interaction partners. In the presence of humans, robots are required to take safety as well as human comfort into account. This paper presents a novel, planning-based approach for social robot navigation. It uses predicted human trajectories and a social cost function to plan collision-free paths that take human comfort into account. It furthermore employs time dependent, kinodynamic path planning to reason about human motion over time and to account for the kinematic and dynamic constraints of a robot. Our approach generates paths that exhibit properties similar to those used in human-human interaction, such as waiting for a human to pass before continuing along an intended path, avoiding getting too close to another human's personal space, and moving out of the way when blocking a human's path. In extensive experiments carried out with real robots and in simulation we demonstrate the performance of our approach.