Bringing Together Human and Robotic Environment Representations - A Pilot Study

Human interaction with a service robot requires a shared representation of the environment for spoken dialogue and task specification where names used for particular locations are depending on personal preferences. A question is how such human oriented models can be tied to the geometric robotic models needed for precise localisation and navigation. We assume that this integration can be based on the information potential users give to a service robot about its working environment. We further believe that this information is best given in an interactive setting (a "guided tour") in this particular environment. This paper presents a pilot study that investigates how humans present a familiar environment to a mobile robot. The study is set up within our concept of human augmented mapping, for which we assume an initial "guided tour" scenario to teach a robot its environment. Results from this pilot study are used to validate a proposed generic environment model for a service robot