Obtaining an object position using multimodal interaction for a service robot

Gestures are useful factor in communication. A gesture has different meanings depending on related objects, situation and so on. Authors aim that a robot recognizes gesture meanings and provides a service considering related objects. Object position is one of the important factors for a robot to recognize a gesture. This paper focused on multimodal interaction between human and a robot to obtain object positions. Proposed system combines gesture recognition and speech recognition as multimodal interaction, and manages the database with object position information through the interaction. The system expects improvement of infrastructure to support cognitive ability of robots.