摘要 本論文將探討模糊滑動控制器於隧道內營建機器手臂的定位及應用於隧道表面的全自動偵測。在本論文控制器的設計上，由於實驗中機器手臂的動態模型相當複雜，故採用模糊控制器做為主要的架構，且結合類似滑動模式控制的觀念來減少控制輸入使設計簡單化，本論文所提出的控制器為一單輸入單輸出的系統，而且只有五條模糊規則，除此之外，此設計也改善了系統暫態響應的上升時間並獲得較佳的抗雜訊能力。另外再結合一灰色預測器做為前饋式補償器來消除系統的超越量。 一旦機器手臂有精準及良好的定位控制之後，將設計完成的控制器結合超音波來實現隧道表面偵測的全自動化，然後記錄偵測到障礙物的中心點位置、長(角度值)、寬(公分)、深(超音波感測值轉換得來的實際值)。透過視窗化軟體，設計出一套完整簡單的使用者操作介面，讓我們可以在做實驗時操作上更加簡單方便。 綜合以上實驗，我們已經突破以往對機器手臂的單軸控制，達到可同時多軸的定位控制，且將其應用於全自動的偵測，最後，我們發現大部分的實驗結果都能達到我們的目標，本論文也將對小部分結果不甚理想的原因提出探討及未來改進的方向。 ABSTRACT A fuzzy sliding controller with a gray predictor for the motion of the robot manipulator is proposed in this thesis. The gray predictor that acts as a feedforward compensator is used to overcome the excessive overshoot of the system transient response. Here, we add the quasi idea of sliding mode control (QSMC) to the fuzzy control. Thus, fuzzy sliding control can reduce the system inputs and fuzzy rules. And the system will be a SISO system and it just has only five fuzzy rules. The process of design will become simpler. Besides, it also has better ability to avoid disturbances and the transient response will be ameliorated in advance when QSMC is used. The idea is employed to the position control of a three-axis robot manipulator. The experimental results show that the proposed controller not only drastically reduces the overshoot, but also maintains a small extent of steady-state error, and has a better rising time. After we can control the motion of manipulator precisely, we use our design to realize the tunnel’s detection by the manipulator automatically. We also design an interface under the windows by Borland C++ Builder to operate the manipulator directly and conveniently.
