부산대학교 도서관

When multiple people are around a communication robot, it is important for the robot to select one proper person whom it talks to. This paper proposes a new communication partner decision system for a robot. This system uses human tracking information from a laser range finder and face detection from cameras to assess one's interest. It can track movements of multiple people by using the laser range finder and choose some persons as candidates for conversation who are standing near the front of the system or approaching to it. Then, the system counts the frequency of face detection as a degree of the interest to a robot for each candidate, and select one person as a first communication partner from the plural candidates under the assumption that a person who watches a robot system most frequently is most interested in it. Some experimental results reveal the validity and effectiveness of our proposed communication partner decision system.