Planning and Realizing Questions in Situated Human-Robot Interaction

Ivana Kruijff-Korbayová; Geert-Jan Kruijff; Miroslav Janicek

In: Proceedings of the AAAI Fall Symposium on Question Generation. AAAI Fall Symposium, November 4-6, Arlington, Virginia, USA, AAAI Publications, 2011.


Robots need to deal with uncertain or incomplete input about the environment. They can thus benefit from interacting with humans to acquire information. When asking a question, a robot needs to say or otherwise signal enough about its belief state and intentions, for the human to clearly understand what it is after. We survey existing work on the forms and meanings of questions in English, concentrating on the issue of how besides eliciting infor- mation from the hearer, a question can simultaneously offer a window into the speaker�s belief state. We propose a formalization based on a notion of common ground, set in a model of situated dialogue as part of collaborative activity. Generating questions starts from agent belief modeling, then forming the intention to request �missing� information or elicit feedback on �uncertain� information from a human, and planning and constructing the surface realization, including syntax and intonation.


kruijff-korbayova.pdf (pdf, 164 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence