本文探討以具有情緒表達特性之機器人其臉部表情特徵演算與變化之機制。本文討論如何藉由二維 情感座標系統分類建立出具有情緒狀態之臉部表情，再定義出各表情變化使用之控制特徵點。如此則可 對不同情緒狀態之間其二維情感座標值計算出表情特徵之變化。為了使情感之表達更為接近一般人物之 特性，此特徵點再經由貝茲曲線運算產生出更為平順化的表情，此與設計之出情感座標系統結合以生成 多樣且細膩之情緒行為輸出。本文最後以LabVIEW 模擬情緒狀態之臉部表情特徵輸出，就一般人對其所 展現之情緒狀態進行認知實驗以測試分析本研究對於機器人以臉部展現其情緒狀態之有效性。

This paper discusses the facial expression generation mechanism of s the social robot . This article discusses how the two-dimensional emotional plane is used in the classification and establishment of facial expressions containing mood information. The algorithm for the expression variation under mood transition is also studied and verified. The smoothness of this transition is vital for the robot to mimic the human communication skill in the time domain history for showing the dynamic characteristics of live mood. The geometric feature control points of robot face are defined. Control points are used to generate the facial expression by the PA coordinate of the emotion coordinate system in the first place. The mood variation can then be mapped to the facial changes directly by the PA parameters. The Bezier curve function is used to generate the geometric shape among control points for more smooth and natural appearance. Finally, LabVIEW simulation of this facial expression and generation model is constructed. The cognitive experiments are conducted. The results demonstrate the effectiveness of this model.