機器人導航系統，是成功的機器人的最重要也最基本的元件之一。一個機器人的導航系統是該機器人能傑出地移動的關鍵。在本篇論文中，我們提出一適用於自主移動機器人的導航系統。我們的導航系統是由基於行為與基於模型這兩種導航系統所混合而成。在我們的系統中，由基於行為子系統來負責低階的即時反應行為，由基於模型子系統來負責做高階的有計畫性的行動。此外，我們的系統能跟無線感測網路來通訊，並可利用無線感測網路的定位技術來幫忙校正機器人的估計位置。當機器人將前往一目的地時，我們的系統將利用基於模型子系統來計算出從機器人到該目的地的路徑。然後，它會把此路徑切分成許多虛擬點，而基於行為子系統將會逐一地接近每一個虛擬點。如果有若干障礙物妨礙到，導航系統將使用我們的避障演算法來閃避這些障礙物並確保機器人能持續地前往目的地。因此，我們的機器人將會正確地抵達目的地。而且，我們使用多執行緒技術來建構我們的導航系統。如此一來，我們的系統便可以同時執行重要的模組們且能更加有效率地利用多核心處理器。根據我們的實驗結果，我們的導航系統在有障礙物的走廊之中能有效地導航機器人，並且可以應用廣泛。 The robotic navigation system is one of the most important and fundamental components of the successful robots. A navigation system of a robot is the key to the excellent motion of the robot. In this thesis, a navigation system for autonomous mobile robot is proposed. Our navigation system is a hybrid of behavior-based and model-based navigation systems. In our system, behavior-based subsystem is in charge of low-level reactive actions, and model-based subsystem is responsible for high-level planned actions. Besides, our system can communicate with wireless sensor network and utilize the localization technology of wireless sensor network to calibrate the estimated position of the robot. When the robot is going to leave for a destination, our system will utilize model-based subsystem to compute a path from the robot to the destination. Then, it divides this path into many virtual points, and the behavior-based subsystem is going to approach each virtual point in turn. If there are some obstacles in the way, the navigation system will use our obstacle avoidance algorithm to avoid these obstacles and keep the robot toward the destination. Therefore, our robot will arrive at the destination correctly. Furthermore, we use multi-thread technology to establish our navigation system. Thus, our system can run important modules concurrently and can utilize the multi-core processor more efficiently. Based on our experimental results, our navigation system can navigate the robot in the passages with obstacles effectively and would be applied extensively.
