具情緒感知人臉機器頭顱之設計與研製

有鑑於目前互動機器人的開發上，大部分著重於機器人互動之動作及行為表現，而無情緒與表情的呈現，故機器人與人互動時，缺少了互動的真實感，猶如一個冷冰冰的機器人。本論文主要是設計一台具有情緒感知的智慧型機器頭顱，此機器頭顱具有四個主要功能，第一、可以流暢的呈現多種基本表情：快樂、難過、生氣、害怕、厭惡、困惑、驚訝與鬼臉，搭配不同顏色之燈光呈獻臉色之變化與簡易的語音回饋，使得表情反應更加生動；第二、機器頭顱對於不同事物有喜好的情緒反應，例如：當機器人看見喜歡的事物時，會表現出開心的表情及語言；看見可怕的事物，則發出尖叫聲音與表情…等；第三、即時追蹤物體的能力，機器頭顱會隨著感興趣的物體，眼睛與脖子做出相對應的轉動；第四、拍打偵測，當機器頭顱受外力拍打時，會有激烈的情緒反應。此研究由兩位同學所合力完成，另一位同學著重於討論機器人頭的視覺與即時追蹤功能。而本篇論文主要是設計與製作一顆具情緒感知的機器頭顱硬體。此機器頭顱使用十一顆AI-1001馬達來牽動十六個可動點，搭配自行設計的機械結構與外觀，可完成八種基本表情呈現，其控制方法上採用馬達角度控制、馬達扭力控制、時間延遲、表情動作規劃、LED燈控制與簡易語音回饋。為了增加機器頭顱的感知能力，在機器人視覺方面，機器頭顱透過無線攝影機擷取即時畫面，利用畫面的RGB資訊來辨識複雜環境中的各種道具，並對於道具有喜好程度之分別；另外，使用AI馬達的旋轉感測器來偵測機器頭顱是否受到拍打。本論文的最大成就在於賦予機器人情緒感知，並增進與人類互動上的真實性與趣味性。

關鍵字

機器頭顱；情感；互動

並列摘要

There have been a lot of studies and designs on the robot development for the interaction between robots and human. However, most of them emphasize the motion interactions. If the motions or reactions of the robot lack emotions and feelings, it just a cold robot with lifeless interaction. Thus, the main objective of this thesis is to design and implement an emotions robot that has emotion expressions to interact with human. The robot has four capabilities. First, he can shows eight different facial expressions including happiness, sadness, anger, fear, disgust, confused, surprised, and grimaced, etc. Second, the robot has different response to the different objects. For example: the robot head will be happy when he sees its favorite object; on the other hand, the robot will scream out if he sees a frightening object. Third, he can achieve the object tracking. Here, the robot will turn his neck and eyes to track the interested object. Finally, the robot can detect whether he is hit suddenly. When the robot is hit by someone or something, he will be angry. This project is accomplished by two students. This thesis discusses the design and hardware implementation of emotional robot head, while the other thesis focuses on the real time object tracking. We use eleven AI-1001 motors to control sixteen degrees of freedom such that the robot head can show eight different emotional expressions. The control strategies include rotation degree control of motors, motion timing control, and expression schemes, LED lights control, and voice feedback control etc. Since there is a wireless camera set up in the robot head, the robot can recognize and track the object according to the extracted images by the camera. Moreover, the encoder of the AI motor can serve as a sensor to detect whether the robot head is hit suddenly, and then the head can present the response for the hitting. We have to emphasize that the main contribution of the thesis is to enable the emotions for the robot head and to increase the interaction between the robot and human.