情感機器人頭之追蹤系統的設計與實現

論文摘要本研究「具有情緒感知的智慧型機器人頭」是由二位同學合力完成，此機器人頭具有四個主要功能，第一、可以流暢的呈現多種基本表情，如：快樂、難過、生氣、害怕、厭惡、困惑、驚訝與鬼臉…等；第二、機器人頭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；第三、即時追蹤物體的能力，機器人頭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，當機器人頭受外力拍打時，會有激烈的情緒的反應。機器人頭的機構中動力部分採用AI-1001馬達，開發軟體為Borland C++ Builder 6，電腦人機介面透過RS-232來進行馬達的角度與扭力控制，搭配自行設計的機械結構來完成機器人頭的主體架構。其控制決策上採用馬達角度控制、馬達扭力控制、時間延遲控制、表情動作規劃、LED燈控制與簡易語音回饋，來實現各種功能。此研究由兩位同學所合力完成，另一位同學主要在設計與製作一顆具情緒感知的機器頭顱。而本論文著重於討論機器人頭的視覺與即時追蹤功能。當目標物進入機器人頭的視覺範圍時，機器人馬上進行影像前置處理，計算出待物體中心點位置，接著使用畫面分割法或模糊控制法，當作機器人頭的移動策略，以控制脖子、眼睛的馬達旋轉，使機器人頭達到追蹤目標物的功能。畫面分割法是將機器人視覺畫面水平、垂直各分為5個區間，待物體落在各個區塊，則賦予不同編號馬達負責轉動追蹤；而模糊控制法則以物體在機器人視覺畫面水平、垂直座標為輸入，輸出則為脖子、眼睛上下及左右的5顆馬達轉動角度，以25條模糊規則實現追蹤目標物的目的。

關鍵字

目標物追蹤；移動策略；模糊控制

並列摘要

Abstract This study is a team work by two students, which designs and makes a robot head which has emotional expression and can interact with people. The robot has four capabilities. First, it is capable of performing different facial expressions, such as happiness, sadness, anger, fear, disgust, confused, surprised, ghost face, etc. Second, the robot has different reactions to different objects. For instance, the robot will be happy when he sees its favorite object. On the other hand, the robot will scream if he sees awful object. Third, he has the capability of object tracking. Here, the robot can move its neck and eyes to track the interested object. Fourth, the robot will be angry if he is hit by someone or something. The movement mechanism of the robot head is achieved by implying AI-1001 motors which are controlled by using software Borland C++ Builder 6. Time delay control, motors control, LED control, and facial expressions regulation are implemented by computer interface with RS-232 connection. The LEDs are employed to emphasize the emotion expressions. In addition, we also add some voice to increase the reaction of the robot. This project is accomplished by two thesises. This thesis focuses on the real-time object tracking and the visual perception of the robot head and the other thesis focuses on the hardware design of emotional robot head. When a desired object was captured by the CCD in the robot head, the system will calculate the position of the object by using image processing technology. The frame partition or fuzzy logic is employed as robot head movement strategy. Based on the movement strategy, we control AI motors to turn around the neck and to rotate eyes of the robot head such that the object tracking is achieved. The frame partition method divides the horizontal and vertical area of robot vision into five block areas. Different rotation degrees of the corresponding motors are set for different areas. Thus, the robot can track the object at each area of robot vision. Furthermore, fuzzy logic method uses 25 fuzzy rules to achieve the object tracking. The position (x,y) of the object as the input of the fuzzy rules and the rotating degree of the corresponding motors is the output. Experiments have shown that the control effect is successful.