本論文建立情感機器人長期互動之情緒、心情與態度六一情感模型（A Six-One Emotion, Mood and Attitude Affective Model for Long-term Interaction, SO-EMA），以量化的方式來表達機器人的各種情緒，並建立一套情感與表情的相對應關係，隨著外部事件刺激與內部情緒之轉變，以及針對不同對象的互動差異性，進一步設計出機器人長期互動的情感演算法。
最後，用實驗結果來證明SO-EMA在長期與人類互動符合人類的情感表現，因此，SO-EMA演算法的建立使得機器人與人類的互動更自然、類似人類的情緒反應且發展更加全面。 ;In this thesis, we build a social affective robot by quantizing robot’s every kind of Emotion and building a corresponding relation of affection and expression. Affection includes Emotion, Mood and Attitude. Along with the transition of internal and external stimulus and the interactive difference regarding to different objects, we design the affection algorithm (SO-EMA) of long-term interaction and behavior improvement.
In SO-EMA, with time and different objects’ even evaluating, robots receive stimulus even by different objects, calculate the weight of quantifiable Affection responds by adjustable personality and Attitude, and combine the present affection to generate the next state of affection. In the process of Affection transition, robot keeps different Attitude to different objects. The next state of Attitude will be generated by interaction time and Affection fluctuating periodically. Thus, in long-term interaction, we can see the differences on the Attitude between different objects and the true affective expression of human in group.
In the end of this thesis, we propose several experimental data to prove that SO-EMA meet the affection expression of interaction, and improve the relationship between human in long-term.
