在不久的將來，各式不同種類、功能的機器人，將會出現在我們的生活環境中，因此如何使機器人利用自身的感測器，在人類的生活環境中執行任務，是一個相當重要的研究課題。 本論文提出了一套能讓人型機器人在操控者下達命令後，自主行走至目標區域的方法。此方法是讓人形機器人在未知的環境空間中，利用感測器，收集機器人所處空間之環境資訊，回傳至系統處理晶片上。透過我們所設計的模糊控制器控制馬達改變角度及利用多層感知機產生步態動作，控制人型機器人在行走中的所有步態動作，使人型機器人能順利的到達所指定的目標區域。我們以嵌入式平台來完成本論文的硬體架構實現，實現系統整合晶片(SOC)的理念。 在所設計的自主行走系統中，我們將本系統安裝在KONDO公司所生產的人型機器KHR-1上，讓人形機器人利用自身的感測器，順利避開障礙物到達所指定的範圍內；也能在斜坡上，藉由自動平衡機制，自動調整身體傾斜角度，避免跌倒發生。 More and more kinds of robots appear in our daily life in recent years. It will be a very challenging task to make robots be able to use their own sensors to execute assignments in our living environments. In this paper, we propose a method which allows a humanoid robot to be able to walk by himself to the destination after the operator gives his or her instructions. The humanoid robot can adjust its motor angles to walk by the use of trained neural networks. It uses an ultrasonic rangefinder to get the environment information and knows how to arrive the destination without bumping to obstacles by use of trained fuzzy controllers. It also can automatically tune the angle of inclination of its body to avoid its falling by the auto-balance mechanism based on a fuzzy controller. We implement all algorithms on a chip, C8051F120. We use the humanoid robot, KHR-1, produced by the KONDO Corporation, to demonstrate the effectiveness of our ideas about an autonomous walking robot system. The humanoid robot incorporated with our proposed autonomous system can use its own sensors to automatically avoid obstacles and then arrive at the destinations; it also can automatically tune the angle of inclination of its body to avoid its falling by the auto-balance mechanism.
