Referent identification process in human-robot multimodal communication

摘要

This paper presents a communication robot that can generate a referent identification conversation with human users. First, we conduct an experiment to collect face-to-face referent identification communication and investigate how the referent is identified by exchanging multiple speech turns between the participants. On the basis of the experimental observations, we implement a communication robot that can manage a referent identification conversation with a user by integrating the linguistic information obtained from speech recognition and the vision information obtained from a robot camera.