無人機前方視野精準輪廓線條跟隨方法之研究

Accurate contour line-following methods for UAV forward view

李恭儀(Lee Gong-Yi)

指導教授 ： 劉吉軒

國立政治大學/理學院/資訊科學系/碩士(2018年)

https://doi.org/10.6814/THE.NCCU.CS.001.2019.B02

摘要

現有的線條跟隨技術主要讓無人機跟隨下方視野的線條，以等高定速向前跟隨，並調整轉向角改變跟隨方向。然而當無人機跟隨前方視野的線條時，調整轉向角會使無人機前方視野的線條消失，而且現有的線條跟隨不會再偵測到跟隨過的線條，但前方視野依然能偵測到跟隨過的線條，需要固定前方視野來決定跟隨方向。而且當無人機改變方向飛行時，定速所產生的移動慣性會使無人機偏離原本的飛行路徑。除此之外，現有的研究亦存在一些問題，如線條的錯誤偵測、沒有固定的方式評估線條跟隨的表現，以及未要求跟隨的精準度。
因此本研究提出二維方向向量機率模型，解決無人機跟隨前方視野的線條時所產生的方向性問題以及避免線條的錯誤偵測影響跟隨。本研究以雙層二維方向向量機率模型，搭配慣性速度抑制方法，能夠抑制無人機改變方向飛行時所產生的移動慣性，使無人機能夠快速且精準的進行線條跟隨。本研究提出兩種評估指標1. 無人機視覺中心位於目標路徑寬度以內之程度以及2. 無人機視覺中心偏移目標路徑寬度以外的位移誤差，評估無人機進行精準的線條跟隨時的表現。
最後本研究透過在真實世界的實驗，驗證提出跟隨方法的可行性、穩定性以及精準性。以先前研究中最能精準跟隨線條的基於向量域的線條跟隨作為基準，本研究所提出的方法經過兩種評估指標進行評估後，皆比基準表現得更好。其中，雙層二維方向向量機率模型搭配慣性速度抑制方法的表現最為突出，該方法具備選擇飛行方向、校正自身位置以及慣性速度抑制的功能，能跟隨複雜的線條。經過真實世界的考驗，本研究提出的方法能實際應用在真實世界上。未來能針對前方視野的線條跟隨研究進行更進一步的改進與延伸，包含了移動時的穩定性、改進位移誤差以及戶外實際應用，如高壓電塔檢測、摩天樓設備安檢等任務。

關鍵字

無人機 ； 自主飛行控制 ； 輪廓線條跟隨 ； 二維方向向量機率模型

並列摘要

Majority of existing line following techniques focused on allowing the drone to follow lines located bottom of the drone’s front view camera. The drone are often in constant speed and changes its following direction by adjusting its steering angle. However, when the drone needs to follow lines located vertically at the center of the front view camera, adjusting the steering angle will make the line disappeared from the drone’s vision. Even though the previously followed line can still be seen in the front view camera, current existing line following techniques cannot detect a previously followed the line, the line needs to be fixed in front of the view to determine the direction of the follow. Moreover, when the drone changes direction, the moving inertia will cause the drone to deviate from the original flight path. In addition, there are still rooms of improvement for existing research, such as error detection of lines, lack of common methods to evaluate the performance of line following and does not include line following accuracy as performance measurement.
Therefore, this study proposes a two-dimensional directional vector probability model to solve the directionality problem caused by the drone following the line of the front view and to avoid error detection of the line. In this study, the two-layer two-dimensional vector probability model and the inertial speed suppression method can suppress the moving inertia generated by the UAV when changing direction, allowing the drone to follow line quickly and accurately. This study also proposes two evaluation indicators to evaluate the performance of the drone for precise line following: 1) The degree of UAV vision center within the width of the target path, 2) The displacement error of the drone vision center from the target path.
Finally, this study verifies the feasibility, stability and accuracy of the proposed method by experimenting in the real world. Based on the most accurate vector-based line following methods in the previous study as the benchmark, using our proposed evaluation methods, the proposed method in this study performs better than the benchmark. Among them, the two-layer two-dimensional direction vector probability model and the inertial speed suppression method are the most prominent, it has the capability of selecting flight direction, correcting its own position and suppressing the inertia speed, and can follow complex lines. After experimenting in the real world, the method proposed in this study can be practically applied in the real world. In future, we can further improve and extend the line following research for front view, including stability during movement, improvement for displacement error and practical applications for outdoor, such as high-voltage tower inspection and security inspection for skyscraper facilities.

並列關鍵字

UAV ； Autonomous flight control ； Contour line following ； Two-Dimensional direction vector probability model

參考文獻

[1]Khan, M. I., Salam, M. A., Afsar, M. R., Huda, M. N., & Mahmud, T. (2016, July). Design, fabrication & performance analysis of an unmanned aerial vehicle. In AIP Conference Proceedings (Vol. 1754, No. 1, p. 060007). AIP Publishing.

[2]Intel Corporation. (n.d.). Retrieved January 9, 2018, from https://click.intel.com/intel-aero-ready-to-fly-drone.html

[3]DJI M200. (n.d.). Retrieved November 23, 2018, from https://www.dji.com/zh-tw/matrice-200-series?site=brandsite&from=nav

[4]Parrot Drones SAS. (n.d.). Retrieved October 2, 2018, from https://www.parrot.com/us/drones/parrot-bebop-2.

[5]Punetha, D., Kumar, N., & Mehta, V. (2013). Development and Applications of Line Following Robot Based Health Care Management System.International Journal of Advanced Research in Computer Engineering & Technology (IJARCET), 2(8), 2446-2450.