擬人化エージェントにおける視聴覚アテンション機能の実現

抄録

For the cognitive interface in user-agent interaction, it is possible to improve the flexibility of interaction if the attention function, which connects visual or acoustical sensors with cognition, was implemented. This paper proposes an approach to sense the location and the acoustic power of speaker, which will be combined with skin color feature in order to draw the attention of agent. In the attentive area, the agent detects user’s face and estimates its orientation so that it can share the user’s attention for the forthcoming communication.