利用FAA*演算法於無人船的即時避碰路徑之規劃

REAL-TIME PATH PLANNING FOR UNMANNED SURFACE VEHICLE BY USING FINITE ANGLE A* ALGORITHM

近年來，無人自走船的相關研究正不斷的進行著，如何降低無人船航行成本成了眾所討論的熱門議題之一。因此，無人船的最短路徑規劃之研究便因應而生，正如火如荼的被探討著。先前研究中的Finite Angle Astar演算法屬於事先規劃，無人船的航行路線需由岸上電腦將路徑規劃好後供給無人船，再依其路徑航行，一旦無人船航駛後，其規劃路徑便無法改變。而本研究中，Real-time FiniteAngle Astar演算法將被提出，使無人船即使於航行中也能任意依需求改變其目標地並規劃出最短之路徑，除此之外，若有障礙物出現也能透過感測器的偵測，將障礙物即時加入地圖中並從新規劃路徑，確保規劃出為最短且安全的路徑。本研究利用衛星空照圖及影像分析技術，將彩色的空照圖轉換為二值化影像供Real-time Finite Angle Astar演算法使用，其所規劃之路徑與空照圖本身同屬WGS座標系，可直接供無人船航行使用。本研究亦設計適用於雙直流推進器的無人船模糊控制器，搭配電子羅盤及GPS模組完成無人船之導航。 Recently, the development of autonomous surface vehicles (ASV) is progressing and is becoming a major focus where lowering the cost and processing time have become critical issues. Hence, shortest path planning for unmanned surface vehicle (USV) is currently being investigated extensively. In this investigation, a Real-time Finite Angle Astar (FAA*) method is proposed to improve the shortcoming of FAA* mentioned in previous studies. The method enables the USV’s route to change immediately. Once unknown obstacles in the planned route are detected, the new shortest path for the USV can be re-planned at once, and then path is assured to be both shortest and the safest. Image analysis is utilized to convert color satellite images into binary images which can be used as the maps of Real-time FAA*. In this article, fuzzy logic navigation is addressed and analyzed. The fuzzy controller of USVs is designed to automatically control two DC brush thrusters. The input variables of the fuzzy controller are provided by the GPS module and the electrical compass. Furthermore, the output variables are used by the Pulse Width Modulation (PWM) technique for the DC thrusters.