Situating Spatial Templates for Human-Robot Interaction