基於視覺導航之自主無人機環繞檢測

Autonomous UAV Surround Inspection based on Visual Navigation

張為超(Chang Wei-Chao)

指導教授 ： 劉吉軒

國立政治大學/理學院/資訊科學系/碩士(2020年)

https://doi.org/10.6814/NCCU202100038

摘要

與在空曠地區的進行的航空影像不同，人造建築物的航空檢視需要無人機進行更複雜的導航。無人機需要以可控的方式向目標物體移動，以獲取結構表面的特寫影像，同時也需要為了自身的安全而避免碰撞到目標建築物。在本文中，我們提出了一項基於視覺導航的人造建築物之自主檢視任務。我們利用SLAM做為視覺定位之基礎，主要針對以圓柱形建築物為路徑之環繞檢視為例，以實際飛行的形式進行了測試。
我們的技術貢獻主要有兩個方面。首先，我們以一個較為完整之任務形式呈現我們的研究，無人機從起飛開始，接著會自主辨識出目標建築物並設定環繞檢視之路徑，在環繞的同時進行實時校正，完成使用者設定之環繞回合數後便會進行返航。其次，我們使用行為樹作為控制體系結構來集成所有功能組件以增強整體之穩定性以及可行性，並在低成本之微型無人機上進行開發。而在現實世界中的實驗表明，無人機可以以一定的成功率執行環繞檢視任務，並且能完整的獲取目標建築物的影像以進行結構檢視。

關鍵字

無人機 ； SLAM ； 行為樹 ； 建築物檢視

並列摘要

Unlike aerial imagery in open fields, aerial inspection on man-made construction requires more complex navigation from drones. The drone needs to move toward target object in a controlled manner in order to acquire close-up views on structure surface, at the same time, avoid collision for its own safety. In this paper, we present a research work on autonomous visual navigation for aerial inspection on man-made construction. In particular, we focus on developing orbital inspection of pole-like objects. We use SLAM as the basis for visual positioning and we test our method in the form of actual flight.
There are two main aspects of our technical contribution. First, we present our research in the form of a relatively complete mission. The drone will automatically identify the target building and set the path for the surround view from the start of take-off, and perform real-time adjustment while orbiting to complete the user-defined surround After the number of rounds, it will return home. Secondly, we use behavior tree as a control architecture to integrate all functional components to enhance the overall stability and feasibility, and develop it on a low-cost UAV. Extensive experiments in a real world scenario have shown that UAV can perform surround building inspection tasks with a certain success rate, and can obtain complete images of target buildings for structural inspection.

並列關鍵字

UAV ； SLAM ； Behavior tree ； Building inspection

參考文獻

[1] International Civil Aviation Organization (ICAO). Unmanned Aircraft Systems (UAS); ICAO: Montreal, QC, Canada, 2011.

[2] C. Stöcker, E. Anette, and K. Pierre, "Measuring gullies by synergetic application of UAV and close range photogrammetry—A case study from Andalusia, Spain." Catena vol. 132, pp. 1-11, 2015.

[3] C. Yuan, Y. M. Zhang and Z. X. Liu, "A survey on technologies for automatic forest fire monitoring detection and fighting using unmanned aerial vehicles and remote sensing techniques", Canadian Journal of Forest Research published on the web 12, March 2015.

[4] H. Aasen, E. Honkavaara, A. Lucieer, and P. Zarco-Tejada, “Quantitative remote sensing at ultra-high resolution with UAV spectroscopy: A review of sensor technology, measurement procedures, and data correction workflows,” Remote Sens., vol. 10, no. 7, p. 1091, 2018.

[5] M. Israel, "A UAV-based roe deer fawn detection system", Proc. Int. Conf. Unmanned Aerial Veh. Geomatics (UAV-g), vol. 38, pp. 1-5, 2011.