Application of AUV Technology in Seafloor Classification and Submarine Cable Route Design

本研究以自行發展、近即時（Real-Time）的聲納資料自動處理、判釋與路線決策技術結合自主式水下載具（Autonomous Underwater Vehicle；AUV），來準確、快速且自動地繪製出待測海域內可行的海纜舖設路由，作為後續細部調查與路徑規劃的重要參考依據，對於未來海纜建設工作的執行與推動有著實質效益上的直接助益。此外，本研究提出了一種相異於傳統聲納資料處理方式的方法與程序，以處理聲納掃描線取代聲納影像分析的概念與建構並應用網格圖方式來替代傳統習慣採用的直觀式特徵物地圖的作法，不僅實現了聲納資料近即時化處理與分析也同時有效的提高了以聲納執行水下載具導航的可靠度。本研究所發展的技術程序分為三個階段：首先使用以卡曼濾波器為基礎所衍生之間接式即時定位與地圖建立的方法，藉由監控觀測性與估測慣性導航儀誤差狀態來回授修正以提高自主式水下載具導航的精確度；接著以聲納掃描線利用佔格模型來建置網格圖，並以樸素貝氏分類器的三個特徵函數來偵測與定義海床目標物的位置，並完成海床型態分類；最後利用機率式路線圖與A星演算法在考量海纜路由上海纜長度、海床型態、轉點角度、海床坡度與高程變化等設計五大決策因子條件下，於計畫的勘測廊道內自行初步劃定一條可能的舖設路徑，以作為後續細部調查與路徑規劃的重要參考依據。This study proposes an automated method for submarine cable route design from sonar scanlines collected by an autonomous underwater vehicle (AUV). Traditionally, the route design is carrying out by experienced surveyors and engineers using seafloor survey data. In the work, an automated classification and route planning method using real time data gathered by an AUV is developed to improve the efficiency for submarine cable construction. Firstly, to improve the accuracy of AUV localization, a linear time-varying equation is applied to describe the motion of an AUV that enables the utilization of observability analysis for maneuvering accuracy. Observability is a property in linear system which related to the initial state and the system outputs. An observability-based maneuvering planner is proposed to enhance the vehicle state estimation. Simulation results demonstrated that unobservable modes can be controlled through maneuvering and will affect the accuracy of vehicle state estimaton. Secondly, a novel idea using sonar scanlines and grid-based maps is introduced to map the seafloor for classification. A probabilistic classifier based on Bayes'' theorem and Naïve assumption is applied to distinguish the types of seafloor. A node map is constructed by probabilistic roadmap and then an A-star algorithm is applied to determine appropriate cable routes on a corridor from the node map. Seafloor classification, bathymetry, steep slope, angle of alter course, and cable length are five factors of the A-star algorithm. A field result of a case of the cable route survey between islands was demonstrated. The planned route using the proposed method is close in range to the one recommend by experts. The proposed cable route design method costs less to the one performed by experienced engineers, and therefore is advantageous for the budget saving of the submarine construction