趙猛 姚凱 楊國棟 何煊 梁自澤 翟永杰
摘? 要: 隨著無人機(jī)應(yīng)用領(lǐng)域的不斷擴(kuò)大,無人機(jī)對周圍環(huán)境感知的需求逐漸加大,激光雷達(dá)技術(shù)的應(yīng)用成為無人機(jī)研究領(lǐng)域的重要發(fā)展趨勢。為解決無人機(jī)在飛行過程中周圍環(huán)境復(fù)雜導(dǎo)致現(xiàn)場定位信號較差的問題,基于三維激光雷達(dá),結(jié)合大疆無人機(jī)的Onboard SDK技術(shù),設(shè)計機(jī)載激光點(diǎn)云采集系統(tǒng)以及激光點(diǎn)云處理算法。為了降低激光采集系統(tǒng)的操作難度,設(shè)計避障算法,在飛行既定路線的同時結(jié)合激光點(diǎn)云進(jìn)行避障飛行。利用激光點(diǎn)云SLAM算法,實(shí)現(xiàn)了無人機(jī)對周圍環(huán)境的三維建模,以達(dá)到無人機(jī)對周圍環(huán)境感知的目的,并通過實(shí)際測試驗證了所提方法的有效性。
關(guān)鍵詞: 無人機(jī); 場景建模; 三維激光雷達(dá); 激光點(diǎn)云采集系統(tǒng); 系統(tǒng)設(shè)計; 避障飛行
中圖分類號: TN958?34? ? ? ? ? ? ? ? ? ? ? ? ? ? ?文獻(xiàn)標(biāo)識碼: A? ? ? ? ? ? ? ? ? ? ? 文章編號: 1004?373X(2020)14?0042?06
UAV scene modelling based on 3D lidar
ZHAO Meng1, YAO Kai2, YANG Guodong3, HE Xuan1, LIANG Zize3, ZHAI Yongjie1
(1. Department of Automation, North China Electric Power University, Baoding 071003, China;
2. Luohe Power Supply Company, State Grid Henan Electric Power Company, Luohe 462000, China;
3. State key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China)
Abstract: With the continuous expansion of the unmanned aerial vehicle (UAV) application field, the demand of UAV for the surrounding environment perception is increasing gradually. The application of lidar technology has become an important development trend in the UAV research field. An airborne laser point cloud acquisition system and a laser point cloud processing algorithm are designed based on 3D lidar and in combination of the Onboard SDK technology of DJI UAV, so as to solve the difficulty of poor field positioning signal caused by the complex surrounding environment of UAV throughout its flight process. The obstacle avoidance algorithm is designed to reduce the operation difficulty of laser acquisition system and avoid obstacles by combining the laser point cloud while flying on the established route. The 3D modelling of the UAV for its surrounding environment is realized by means of the laser?point cloud SLAM algorithm to achieve the purpose of UAV′s perception of the surrounding environment. The effectiveness of the proposed method is verified by the actual test.
Keywords: UAV; scene modeling; 3D lidar; laser point cloud acquisition system; system design; obstacle avoidance flight
0? 引? 言
近年來,無人機(jī)技術(shù)的快速發(fā)展使其在工業(yè)領(lǐng)域得到了廣泛的使用和推廣,但是現(xiàn)在的無人機(jī)操控基本停留在手動控制的階段[1],對操作人員的要求門檻較高。在環(huán)境建模方面,現(xiàn)階段多采用雙目視覺攝像頭或者深度相機(jī)進(jìn)行建模,對環(huán)境的敏感性較高,且對硬件的處理器能力要求較高,激光雷達(dá)建模則擁有精度較高、對計算量小的優(yōu)勢[2]。激光雷達(dá)作為無人機(jī)自主飛行的重要傳感器[3],之前由于自身重量和成本的問題難以在小型無人機(jī)上得到應(yīng)用和推廣,近年來隨著激光雷達(dá)技術(shù)的不斷突破,機(jī)載激光雷達(dá)的質(zhì)量也降到了1 kg以下[4],使得3D雷達(dá)應(yīng)用到無人機(jī)上成為可能。同步定位與建圖(Simultaneous? Localization and Mapping,SLAM)技術(shù)是機(jī)器人視覺的一個重要研究內(nèi)容[5],采用此技術(shù)可以通過無人機(jī)自身搭載的激光雷達(dá)等傳感器實(shí)現(xiàn)無人機(jī)的環(huán)境感知,結(jié)合IMU慣性測量元件,無人機(jī)可以估計出自身的定位,進(jìn)而根據(jù)之前的位置實(shí)現(xiàn)建圖[6]。本文通過機(jī)載三維激光雷達(dá)對環(huán)境數(shù)據(jù)進(jìn)行采集處理,建立起點(diǎn)云地圖,實(shí)現(xiàn)無人機(jī)周圍環(huán)境感知,通過障礙物檢測實(shí)現(xiàn)避障功能,以提高無人機(jī)在既定路線飛行的自主能力。