自主式水下載具利用單眼視覺之序列式蒙特卡羅定位研究

Sequential Monte Carlo Localization for Autonomous Underwater Vehicle by Using Monocular Vision

陳漢穎(Han-Ying Chen)

指導教授 ： 郭振華

國立臺灣大學/工學院/工程科學及海洋工程學研究所/碩士(2015年)

https://doi.org/10.6342/NTU.2015.00015

摘要

影響自主式水下載具的關鍵能力在於定位及導航。載具追蹤目標物，以及路徑控制的目的是準確辨識目標物與載具的距離和角度關係，並且透過定位結果與理想路徑的比較後可以得到較小的誤差。本文利用序列式蒙特卡羅定位演算法做為研究的方法，以水池中一組人工裝設的環境為驗證場域。首先，自主式水下載具使用單眼攝影機、電子羅盤和加速度計收集位置的量測資訊，並且建立被觀測物體與單眼視覺關係，及利用色彩空間演算法和邊緣偵測找出目標物在影像中的位置與寬度，以得到載具視訊攝影機與已知目標物間的距離及角度關係，並且整合載具的運動模型，以供自主式水下載具在已知環境下作序列式蒙特卡羅定位。最後，本文利用已知水池環境來驗證此導航法之可行性。

關鍵字

自主式水下載具、水下導航、單眼視覺、序列式蒙特卡羅定位演算法

並列摘要

Localization and navigation are two crucial abilities for autonomous underwater vehicle (AUV) to track target and avoid obstacles in an underwater environment. In this work, an underwater environment was arranged for an AUV to accurately identify the relative distance and angular relationship between the AUV and the target; furthermore, the AUV was commanded to track trajectories using onboard vision. Sequential Monte Carlo localization algorithm is applied for the localization algorithm. The AUV acquires environmental and state information using a monocular camera, an electronic compass, and accelerometers for localization and navigation. While establishing the relationship between observed target and monocular camera by using image color space and edge detection to identify position, width and orientation in the reference coordinate of the target. The Sequential Monte Carlo localization in known underwater environment is then constructed by integrating vehicle motion model. The tank site was used as an example to verify the feasibility of the proposed localization and navigation method.

並列關鍵字

autonomous underwater vehicle, underwater navigation, monocular vision, Sequential Monte Carlo Localization algorithm

參考文獻

[1] C. Lee-Johnson and D. Carnegie, “Mobile robot navigation modulated by artificial emotions,” IEEE Trans. Syst., Man, Cybern. B, Cybern., vol. 40, no. 2, pp. 469-480, Apr. 2010.

[2] J.-K. Yoo and J.-H. Kim, “Fuzzy integral-based gaze control architecture incorporated with modified-univector field-based navigation for humanoid robots,” IEEE Trans. Syst., Man, Cybern. B, Cybern., vol. 42, no. 1, pp. 125–139, Feb. 2012.

[3] E. Menegatti, A. Pretto, A. Scarpa, and E. Pagello, “Omnidirectional vision scan matching for robot localization in dynamic environments,” IEEE Trans. Robot., vol. 22, no. 3, pp. 523–535, Jun. 2006.

[4] J. Kramer and A. Kandel, “Robust small robot localization from highly uncertain sensors,” IEEE Trans. Syst.,Man, Cybern. C, Appl. Rev., vol. 41, no. 4, pp. 509–519, Jul. 2011.

[5] I. Shimshoni, “On mobile robot localization from landmark bearings,” IEEE Trans. Robot. Autom., vol. 18, no. 6, pp. 971–976, Dec. 2002.