Heuristic rules for human-robot interaction based on principles from linguistics - Asking for directions

Robots that are to assist humans in flexible and versatile ways, will not always possess all the information required to fulfill their task. Therefore robotic systems have to be able to retrieve information through natural language communication with humans. Natural language is often vague or ambiguous and thus hard to interpret by technical systems. To enable robotic systems to interpret natural language expressions correctly it is necessary to include findings from human-human communication into the dialog systems of robots. In this work the field of communication topics between human and robot is confined to the communication about space, more specifically to a robot asking a human for directions. This paper gives insight on theories from linguistics research focussing on asking for and giving directions. From these theories 10 heuristic rules for human-robot interaction are deduced, where 4 of the rules apply even to systems that are not able to communicate through natural language. Additionally a first experiment where a robot used the 4 basic heuristic rules to successfully ask passers-by for directions and find the way to an unknown goal location is presented.