智慧型機器人之基於地圖特徵節點的動態路徑規劃與模糊推論型導航系統

Abstract

本論文主旨在設計並實現一套應用於智慧型機器人之動態路徑規劃與導航系統，並透過網路整合手持裝置以提供於室內環境中的使用者服務。路徑規劃可分為全域及區域路徑規劃，全域路徑規劃基於已知環境及其資訊進行最短路徑規劃，可得視野較廣之路徑結果；區域路徑規劃則於未知環境中依感測器回傳之即時環境資訊進行路徑規劃，其路徑結果具有即時反應環境之特性。全域路徑規劃中，如何從已知地圖得知可用之路徑規劃資訊卻顯少論文提及。因此本論文提出一基於地圖特徵節點之動態路徑規劃系統，此系統藉已知環境地圖可演算出代表地圖資訊的特徵節點及其連結關係，經此資訊透過最短路徑演算法計算出最短路徑，並搭配自行開發的模糊推論型導航系統以完成動態路徑規劃，達成同時兼具廣視野及即時反應之導航系統。基於智慧型機器人上之各式感測器偵測環境及演算法推論，模糊推論型導航系統整合各式行為如：死巷逃脫、窄巷行走、閃避障礙、沿牆行走及終點追尋等五項行為，並搭配動態路徑規劃以達成智慧型機器人之導航。最後並透過所設計之機器人服務驗證其結果。

In this study, we aim to design and implement a dynamic path planning and navigation system which can be applied in intelligent robots, and the system offers services for users through integrates handheld devices which connect to the network. The system of path planning can divided in to global and local path planning. Global path planning finds the shortest path with a global view based on a given environment. In the other hand, local path planning is based on the real time information acquired from sensors in an uncertain environment, and it bears the characteristic of real time response of environment. In global path planning, few papers mention how to obtain useful information from maps for path planning. Consequently, this study will bring up a graphic feature node based dynamic path planning system, and with this system we can get feature nodes and their linkage within known environment. Based on this information, the system can obtain the shortest path by a shortest path algorithm. Moreover, the system can cooperate with fuzzy based navigation to accomplish dynamic path planning which has both global view and real time response. According to various kinds of sensors and inference of the algorithm, fuzzy based navigation system fusion behaviors such as impasse escape, moving in a narrow lane, obstacle avoidance, wall following and goal seeking. The navigation cooperates with dynamic path planning to accomplish navigating the robot. Finally, we will examine experimental results by the robot services.