由於自動控制與通訊技術的突破，讓無人自走車的應用範圍更廣、功能更多，除了使用於工廠之外，依其追蹤的原理也時常發現被應用於導盲機器人、探險機器人和救災機器人等，成功地顯示出無人自走車已朝着人性化、智慧化和生活化的趨勢發展。本論文主要考慮設計智慧型控制演算法使得無人自走車能在既定的黑色路徑上行走且不偏離，共提出了四種不同的導引控制方法來進行比較，分別有(1)模糊控制、(2)模糊滑動模式控制、(3)自我學習模糊控制、(4)自我學習模糊滑動模式控制。在控制法則上，因模糊控制的設計方法完全憑藉專家經驗進行設計，可以完全不需得知任何的受控系統動態模式，不少研究顯示模糊控制可獲得不錯的響應結果，但是，模糊控制時常因為過多的模糊規則數造成實作上的困難。為了減少實作上的複雜度，模糊滑動模式控制利用滑動表面的提出來有效地減少所需之規則數。進一步的，本論文提出了自我學習模糊控制與自我學習模糊滑動模式控制，因其擁有線上即時修正控制器參數之優點，對於無人自走車尋軌控制能提供更佳控制效果。最後，本論文分別考慮兩輪差動和全向輪兩種不同驅動方式之無人自走車設計，利用光感測器不同電壓值來判斷自走車偏移量大小，並依據所設計之控制法則導引自走車行走並在ARM Cortex M0微控制器上設計實現，經由實際實驗結果發現所提出之方法均可以有效且精確地導引無人自走車在既定的路徑上移動。 This thesis proposes four intelligent guidance laws for a wheeled mobile which the mobile can follow a black line. The proposed four algorithms include a fuzzy control, a fuzzy sliding-mode control, a self-learning fuzzy control and a self-learning fuzzy sliding-mode control. It is known that a fuzzy control can achieve favorable control performance based on human knowledge. But, the huge number of fuzzy rules makes that the fuzzy control is not easily to be implemented in real applications. To attack this problem, the fuzzy sliding-mode control using a sliding surface is proposed to reduce the number of the fuzzy rules. However, the fuzzy rules base of the fuzzy control and the fuzzy sliding-mode control should be designed by many trials and error tuning processes. Further, this thesis proposes a self-learning scheme for both of the fuzzy control and fuzzy sliding-mode control which are termed as self-learning fuzzy control and self-learning fuzzy sliding-mode control. Based on the self-learning scheme, the fuzzy rules base can online tune by itself to achieve favorable control performance. Finally, the proposed four intelligent methods are implemented in ARM Cortex M0 microcontroller and a two-wheeled mobile and a three-wheeled mobile are applied as an example study. Some experimental results show that the proposed four intelligent guidance laws for a wheeled mobile can track a black line well.
