Using spatial context knowledge in gesture recognition for commanding a domestic service robot

The use of human gestures for interacting with robot systems in domestic environments is investigated. Special attention is paid to the recognition of the user's intent behind a gestural action. The main advantage of our approach is that the human's movement together with valuable information extracted from a spatial scene representation are directly considered while trying to uncover the intention behind a human's gesture. To uncover the intention of dynamic human gestures we use continuous density hidden Markov models. As an application example, instructing a domestic service robot is considered. An event driven control architecture permits easy context switching and meets the demands of an interactive robot assistant.