感性工程-人工表情之研發

在人因智慧工程中，類人機器人已成為現今蓬勃發展的趨勢。從早期的運動穩定性、輕量化、小型化，至今已能解決許許多多包括民生、醫療、科技等問題；然而人體中尤屬臉部的設計最難具體表達，為求生動、更具逼真性，已不僅止於在人臉辨識方面已有很大的進步。本論文透過感性工學的建構，其中包含心理學層面的認知探討與實際人體生理結構。為達成人因工學的感知量化，更進一步地追根人體臉部的醫學結構，係由神經牽動進而控制表情的表情肌(屬第七腦神經)，造成皮膚拉扯以達成表情訊息的傳達。利用類神經網路，將其控制單元(表情肌)予以量化後，透過動作電位來代表肌絲滑動原理的肌肉變形，如同腳踩油門使汽車前進。其中肌肉的運動係由動作電位全無全有定律，正如數學上的二值化概念，進而探討歸類出表徵，以完成人體臉部五官的表情模擬。在未來，期盼可以親眼見到機器人的表情更加生動、自然且豐富，能夠為世界帶來進步動力，而不再是面無生氣的模樣或只是單純地為人類服務。

關鍵字

感性工學；認知科學；表情肌；動作電位；類神經網路

並列摘要

Stability, weight reducing, and miniaturization have become primary interests in intelligence engineering of robot research. Those three subjects relate to many problems include people''s livelihood, medical treatment, science and technology, etc. Although the design method of robot face designing is most difficult particularly to express, it has already improved greatly to request vividly and true to life. In the study, the purpose of constructing a kansei engineering structure is to understand the correlation between psychological cognition and physiological structure. Some techniques in intelligence engineering enable us to quantize people''s feeling. Facial muscles belong to the interior structure of human can cause drag hence change the skin shape. The facial expression is controlled by the information from the seventh cranial nerve. Stepping on throttle and makes automobile advance is similar to the motion potential of the Sliding Filament Theory. The displacement of the muscle result from the All or None Principle is the same concepts as of the hardlims function in the mathematics. Such that this theory could sorts out the characteristics to complete the simulation of facial expression by using neural network. In the future, the robot no longer will have a lifeless look or simply serve the mankind’s need. We expect the expression be more vivid, nature, rich and colorful. The robot shall provide progressive power to the world.