自主式水下載具在紊亂磁場中整合電子羅盤與已知環境之視覺導航研究

Abstract

本研究描述自主式水下載具利用雙眼視覺、電子羅盤以及加速度計在一個磁場紊亂與已知特徵物的水下環境中的定位演算法。在實驗場地由於電子羅盤會會受到環境的干擾導致航向角的誤差變化不穩定，本文利用量測磁場分量的基本性質透過幾何關係修正受干擾的航向角值。自主式水下載具可利用修正後的角度值做角度控制以穩定地搜尋水下已知特徵物，再建立雙眼視覺來觀測載具本身與已知特徵物間的相對距離與角度當作觀測資訊；另外整合修正後的航向角與加速度計的資料去預估載具的運動模型，架構出在磁場紊亂的環境下自主式水下載具的延伸型卡曼濾波器定位演算法。最後，本論文展示在兩個不同場地下的實驗數據，以驗證此方法之可行性。

This work describes the localization for a autonomous underwater vehicle by utilization of stereo vision camera, a digital electronic compass, an accelerometer in known underwater environment subject to magnetic perturbations. In the experimental place, the compass could be disturbed so the heading angles differences of the vehicle are changing unstably. The proposed method of compensating the magnetic components is through the geometric relationship of magnetic components to correct the disturbed heading angles, therefore, the vehicle could control angle steadily by modified headings to search the landmarks which are known in their position. Establishing the stereo vision detects the relative distance and orientation between the known landmarks and the vehicle, then combining the information from accelerometers to perform extended Kalman filter localization algorithm to achieve autonomous localization in the magnetic anomalies environment. Finally, the experimental data in two different test tanks that are subject to strong magnetic anomalies verify the effectiveness of propose method of compensating the heading errors.