趙太飛,高 鵬,史海泉,李星善
蜂群無人機編隊內無線紫外光協(xié)作避讓算法
趙太飛1,3*,高 鵬1,3,史海泉1,3,李星善2
1西安理工大學自動化與信息工程學院,陜西 西安 710048;2湖北航天技術研究院總體設計所,湖北 武漢 430040;3陜西省智能協(xié)同網(wǎng)絡軍民共建重點實驗室,陜西 西安 710048
在戰(zhàn)場復雜電磁環(huán)境下,保證蜂群無人機編隊機間飛行安全和編隊內可靠通信尤為重要。本文提出一種蜂群無人機編隊內無線紫外光協(xié)作避讓算法,結合無線紫外光覆蓋特點設計紫外虛擬圍欄避讓策略,基于增強矢量場直方圖法針對無人機在避讓時的運動狀態(tài)的代價函數(shù)進行改進,采用無跡卡爾曼預測器預測鄰近無人機的飛行狀態(tài)。在兩種預測場景下的避讓仿真中,結果表明,與增強矢量場直方圖法進行對比,本文算法的整體運動軌跡平滑,局部避讓時無明顯抖動,避讓路徑總長度平均減少3.46%,總耗時平均減小18.94%,驗證了蜂群無人機編隊內無線紫外光協(xié)作避讓算法的有效性。
蜂群無人機;無線紫外光;虛擬圍欄;協(xié)作避讓;軌跡預測;增強矢量場直方圖法
蜂群無人機編隊是由大量載荷不同、類型不同的無人機組成,根據(jù)戰(zhàn)場環(huán)境調整編隊內機群數(shù)量及隊形以便執(zhí)行隱秘偵察,重點突防等作戰(zhàn)任務。蜂群無人機編隊存在無人機機間密度大,對環(huán)境信息實時性要求高的特點[1]。由于各類型的電磁干擾無處不在,特別是電子對抗過程中無人機編隊需要保持無線電靜默以降低暴露風險,而無線“日盲”紫外光通信正好能滿足這種通信方式的需求,其優(yōu)勢主要有背景噪聲小、抗電磁干擾能力強、非直視通信、低功耗、高集成度、易于機載[2]。因此,采用“日盲”紫外光協(xié)作無人機編隊飛行能為無人機編隊在強電磁干擾環(huán)境中順利執(zhí)行任務提供有效保障。
路徑規(guī)劃是蜂群無人機編隊順利完成任務的前提,分為全局路徑規(guī)劃[3-7]和局部避讓算法[8-10]。全局路徑規(guī)劃在已知環(huán)境信息的前提下通過各類算法實現(xiàn)規(guī)劃,其優(yōu)勢在于路徑平滑,避讓效果好,缺點是不能適用于實時性高的場景,文獻[5-7]中利用移動物體的運動狀態(tài)預測很好地實現(xiàn)了全局路徑規(guī)劃。局部避讓算法通過一定避讓條件實現(xiàn)在線路徑規(guī)劃,其優(yōu)勢在于適用動態(tài)場景,但是其存在局部極小和路徑抖動等缺點,較其他局部避讓算法而言,增強矢量場直方圖法易于實現(xiàn),局部避讓效果好,得到了廣泛的應用。文獻[11-13]通過結合全局路徑規(guī)劃算法和局部避讓算法實現(xiàn)了在動態(tài)場景下的高魯棒性的路徑規(guī)劃,大大減小了局部極小、路徑抖動等問題,也克服了全局路徑規(guī)劃算法動態(tài)場景適應性的問題。本文主要針對基于無線紫外光通信的蜂群無人機編隊,利用無線紫外光構建無人機安全范圍內的環(huán)境直方圖來避讓和預警該區(qū)域內的鄰近無人機,實現(xiàn)高魯棒性、實時性更好無人機編隊內的機間避讓。
1) 安全飛行:通過判斷,在下一個運動周期鄰近無人機處在鏈路建立區(qū)內并且正在遠離,此時判定鄰近無人機為安全飛行狀態(tài)。
2) 一般危險:通過判斷在下一個運動周期鄰近無人機處于通信區(qū)內,但(+2)的周期內處于預警區(qū),標記該無人機為潛在危險。
3) 緊急避讓:通過判斷在下一個運動周期鄰近無人機處于預警區(qū)內,當前時刻必須執(zhí)行局部避讓。緊急避讓狀態(tài)下,通過最佳方向采樣和最佳速度采樣提供局部避讓路徑,再通過篩選出代價最小的避讓路徑,最終實現(xiàn)了機間自主避讓。
圖1 無線紫外光虛擬圍欄模型
當無人機之間互相接收到波長為2的光信號時,雙方開始通信;當無人機之間互相接收到波長為1的光信號時,無人機之間開始執(zhí)行自保程序。
圖2 無人機運動模型
Fig. 2 UAV motion model
通過轉換關系可求得在大地坐標系下任意無人機運動到點時的位置信息為
無人機運動時的速度和位移可由以下公式表示:
其中:
圖3 無線紫外光虛擬圍欄直方圖
Fig. 3 Wireless ultraviolet virtual fence histogram
虛擬圍欄直方圖本質上是一個無線紫外光虛擬圍欄覆蓋范圍內的可飛行區(qū)域集合,需要通過一些約束條件篩選出適合飛行的區(qū)域。原有算法通過設定可飛行區(qū)域的邊界值,比較二者差值與閾值之間的關系篩選最佳運動方向,但是對于無人機來說,這種選擇運動方向存在一定局限性,下一時刻對當前時刻的影響并未考慮,為此,需要通過預測鄰近無人機在下一時刻的飛行狀態(tài),結合該狀態(tài)選擇當前時刻無人機的運動方向。而無人機系統(tǒng)是非常復雜的非線性系統(tǒng)[17],并且利用無線紫外光設備提供無人機機間距離、方位角時也存在誤差。為了提高避讓的精準度,本文將利用無跡卡爾曼預測器預測鄰近無人機的飛行軌跡及運動狀態(tài)以便于運動方向的選擇。
設無人機作勻變速運動,由于無線紫外光設備及運動系統(tǒng)自身均含高斯白噪聲,則運動狀態(tài)方程()和觀測方程()[18]可表示為
式中:()表示無人機系統(tǒng)所包含的高斯白噪聲,其具有協(xié)方差陣。()表示觀測狀態(tài)下的高斯白噪聲,其具有協(xié)方差陣。
圖4 基于無線紫外光的避讓流程圖
避讓算法參數(shù)如表1所示。無人機狀態(tài)預測時系統(tǒng)噪聲()具有協(xié)方差,()具有協(xié)方差陣,分別如下:
()和()二者不相關,采樣次數(shù)=50次,采樣時間=1 s。表2所示為各個運動狀態(tài)預測初始參數(shù)。
在算法對比中,矢量場直方圖避讓算法(VFH+)為局部避讓算法,故存在局部極小的問題,并且原算法的路徑鋸齒程度明顯,路徑不平滑。增強矢量場直方圖法(VFH*)為利用A*算法全局搜索關鍵避讓信息,局部避讓采用矢量場直方圖的避讓算法,該算法雖利用A*算法獲取了全局地圖信息,但是對于無人機蜂群這類高動態(tài)應用場景存在環(huán)境信息更新不及時影響避讓效果等問題?;诖耍岢隽颂摂M圍欄避讓算法(UAVF),本算法為考慮當前運動物體的運動狀態(tài)和運動物體運動狀態(tài)預測的局部避讓算法。由于VFH+只適合于靜態(tài)障礙物的局部避讓,在此將軌跡預測后的位置狀態(tài)離線顯示在地圖中,查看其避讓軌跡狀態(tài)。在場景一下,選取了近前30 s的軌跡,因為會遇發(fā)生在前30 s內。在場景二下,選取了近前50 s的軌跡。圖5(a)為場景一中局部避讓軌跡圖,圖5(b)為場景一中避讓軌跡局部放大圖,圖6(a)場景二中局部避讓軌跡圖,圖6(b)為場景二中避讓軌跡局部放大圖。
表1 避讓算法參數(shù)
表2 運動狀態(tài)初始參數(shù)
圖5 (a) 場景一中局部避讓軌跡;(b) 場景一中避讓軌跡局部放大
圖6 (a) 場景二中局部避讓軌跡;(b) 場景二中避讓軌跡局部放大圖
從圖5(a)中可以觀察出,三類算法均可安全完成局部避讓,并最終到達目的地。從圖5(b)中可以看出,由于UAVF考慮了自身運動速度及在避讓時的機間距離的冗余,局部避讓時路徑平滑且轉向平緩,機間距離保持良好,無明顯軌跡抖動。VFH+由于未能提前獲知運動物體的運動狀態(tài),選擇從障礙物的下一個前進方向避讓,在實際情況中,極有可能在會遇時出現(xiàn)碰撞,而且避讓時出現(xiàn)了明顯的抖動。VFH*雖然提前全局搜索適合路徑,并在局部完成避讓,但是從開始搜索路徑到避讓,局部路徑較長,有轉向角度。仿真中,VFH*算法共耗時36.01 s,路徑總長度570 m;VFH+算法路徑總長度529 m,共耗時31.45 s;UAVF算法共耗時29.23 s,路徑總長度398 m。相比VFH*算法,總路徑長度減少3.02%,總耗時減少18.82%。
圖6(a)表明,三類算法均可安全完成局部避讓,并最終到達目的地。從圖6(b)可以觀察出,由于VFH+在離線規(guī)劃中未能找到合適避讓方向,故從出發(fā)開始選擇繞過最遠端物體到達終點,總耗時費42.97 s,路徑總長度860 m。VFH*和UAVF均能很好地完成局部避讓,但相比VFH*,UAVF軌跡平滑,轉向角度較小。VFH*總耗時47.48 s,路徑長度離794 m;UAVF總耗時38.43 s,路徑總長度763 m。與VFH*相比,UAVF耗時減少19.06%,路徑總長度減少3.90%
圖7與圖8中的無人機真實軌跡和預測軌跡均為Matlab仿真所得。圖7(a)為場景一中的預測軌跡圖,從圖中可以看出,真實值與預測值整體擬合度較高,但是局部相對誤差依然存在。圖7(b),7(c)為場景一中的預測軌跡局部放大圖,放大比例基本相同,順序為從左上到右下。圖7(b)是預測剛開始時局部放大,可以看出由于采用次數(shù)少,相對誤差比較明顯。圖7(c)預測次數(shù)在25~30左右,相對誤差減少。綜合兩幅圖可以分析出,由于預測次數(shù)的增加,相對誤差在逐漸減小,最后趨于穩(wěn)定。在場景一的預測中,相對距離誤差最大不超過6.83 m,相對速度誤差最大不超過1.88 m/s,相對加速度誤差最大不超過0.17 m/s2。
圖8(a)為場景二中的預測軌跡圖,從圖中可以看出,真實值與預測值整體擬合度較高,但是局部相對誤差明顯。圖8(b),8(c)為場景二中的軌跡局部放大圖,放大比例基本相同,順序為左下到右上。圖8(b)是預測剛開始時的局部放大圖,可以看出,由于剛開始預測,相對誤差非常明顯,圖8(c)是預測次數(shù)在15~25左右,相對誤差減少幅度大。綜合兩幅局部放大圖可以分析出,隨著預測次數(shù)的增加,相對誤差在減小,最后趨于穩(wěn)定。在場景二的預測中,相對距離誤差最大不超過8.19 m,相對速度誤差最大不超過0.82 m/s,相對加速度誤差最大不超過0.11 m/s2。兩次狀態(tài)下的相對誤差平均值如表3所示。
本算法考慮了蜂群無人機編隊內無線紫外光隱秘通信的覆蓋特點,提出了無線紫外光虛擬圍欄避讓策略?;趥鹘y(tǒng)增強矢量場直方圖法,通過增加速度采樣改進代價函數(shù),結合無跡卡爾曼預測器預測編隊內其他無人機運動狀態(tài),實現(xiàn)了蜂群無人機編隊飛行時的協(xié)作避讓。仿真結果表明,與傳統(tǒng)算法相比,本算法在兩種場景中的避讓總路徑長度平均減少3.46%,總耗時平均減小18.94%。該算法能夠在未獲取全局地圖的情況下,通過無線紫外光設備及無人機運動狀態(tài)預測實現(xiàn)協(xié)作避讓。
圖7 (a) 場景一中的預測軌跡;(b) 第一次局部放大;(c) 第二次局部放大
圖8 (a)場景二中的預測軌跡;(b)第一次局部放大;(c) 第二次局部放大
表3 相對誤差平均值
[1] Yu L, Wei P, Ma Z L,. Analysis on the development of foreign army anti-bee colony drone technology[J]., 2017(12): 26–30.
于力, 魏平, 馬振利, 等. 外軍反蜂群無人機技術發(fā)展分析[J]. 飛航導彈, 2017(12): 26–30.
[2] Luo C. The study of signal processing and design of base band system for non-line-of-sight optical communication[D]. Beijing: University of Chinese Academy of Sciences, 2011.
羅暢. 非視距光通信信號處理研究與基帶系統(tǒng)設計[D]. 北京: 中國科學院研究生院, 2011.
[3] Tan Y Y, Li Y, Zhou J,. Path replanning approach for UAV based on A* algorithm in complex environment[J]., 2017, 39(6): 1268–1273.
譚雁英, 李洋, 周軍, 等. 復雜環(huán)境下基于A*算法的無人機路徑再規(guī)劃[J]. 系統(tǒng)工程與電子技術, 2017, 39(6): 1268–1273.
[4] Mac T T, Copot C, Tran D T,. A hierarchical global path planning approach for mobile robots based on multi-objective particle swarm optimization[J]., 2017, 59: 68–76.
[5] Roberge V, Tarbouchi M, Labonte G. Comparison of parallel genetic algorithm and particle swarm optimization for real-time UAV path planning[J]., 2013, 9(1): 132–141.
[6] Wu Z Y, Li J H, Zuo J M,. Path planning of UAVs based on collision probability and Kalman filter[J]., 2018, 6: 34237–34245.
[7] Chen Y B, Luo G C, Mei Y S,. UAV path planning using artificial potential field method updated by optimal control theory[J]., 2016, 47(6): 1407–1420.
[8] Qu P R, Xue J R, Ma L,. A constrained VFH algorithm for motion planning of autonomous vehicles[C]//, 2015.
[9] Babinec A, Duchoň F, Dekan M,. VFH*TDT (VFH* with time dependent tree): a new laser rangefinder based obstacle avoidance method designed for environment with non-static obstacles[J]., 2014, 62(8): 1098–1115.
[10] Zhang Y Y, Qu D, Ke J,. Dynamic obstacle avoidance for USV based on velocity obstacle and dynamic window method[J]., 2017, 23(1): 1–16.
張洋洋, 瞿棟, 柯俊, 等. 基于速度障礙法和動態(tài)窗口法的無人水面艇動態(tài)避障[J]. 上海大學學報(自然科學版), 2017, 23(1): 1–16.
[11] Ji J, Khajepour A, Melek W W,. Path planning and tracking for vehicle collision avoidance based on model predictive control with multiconstraints[J]., 2017, 66(2): 952–964.
[12] Cui R X, Li Y, Yan W S. Mutual information-based multi-AUV path planning for scalar field sampling using multidimensional RRT[J]., 2016, 46(7): 993–1004.
[13] da Silva Arantes J, da Silva Arantes M, Toledo C F M,. Heuristic and genetic algorithm approaches for UAV path planning under critical situation[J]., 2017, 26(1): 1760008.
[14] Zhao T F, Yu X X, Bao H,. Ranging and positioning method using wireless solar blind ultraviolet[J]., 2017, 25(9): 2324–2332.
趙太飛, 余敘敘, 包鶴, 等. 無線日盲紫外光測距定位方法[J]. 光學精密工程, 2017, 25(9): 2324–2332.
[15] Zhang Y L, Zhao C P, Yan H. Path planning for unmanned aerial vehicles using improved bidirectional A* and vector field histogram algorithm[J]., 2019, 19(4): 179–184.
張亞蘭, 趙成萍, 嚴華. 基于改進雙向A*和向量場直方圖算法的無人機航路規(guī)劃[J]. 科學技術與工程, 2019, 19(4): 179–184.
[16] Sun Y Z, Xiao S D, Pan S F,. VFH algorithm for obstacle avoidance based on Kalman Filter[J]., 2015, 29(4): 373–377, 398.
孫揚智, 肖世德, 潘紹飛, 等. 融合卡爾曼濾波的VFH避障算法[J]. 江蘇科技大學學報(自然科學版), 2015, 29(4): 373–377, 398.
[17] Dai R, Cochran J, Jr. Path planning and state estimation for unmanned aerial vehicles in hostile environments[J]., 2010, 33(2): 595–601.
[18] Guan Y Z, Song C L, Dong H J. Path planning of the free-floating manipulator for capturing a moving target[J]., 2017, 39(6): 803–811.
關英姿, 宋春林, 董惠娟. 空間自由漂浮機器人對運動目標抓捕的路徑規(guī)劃[J]. 機器人, 2017, 39(6): 803–811.
An algorithm for the bee colony drone to use wireless ultraviolet for cooperative obstacle avoidance
Zhao Taifei1,3*, Gao Peng1,3, Shi Haiquan1,3, Li Xingshan2
1Faculty of Automation and Information Engineering, Xi¢an University of Technology, Xi¢an, Shaanxi 710048, China;2General Design Institute of Hubei Academy of Aerospace Technology, Wuhan, Hubei 430040, China;3Shanxi Civil-Military Integration Key Laboratory of Intelligence Collaborative Networks, Xi¢an, Shaanxi 710048, China
Wireless UV virtual fence model
Overview:Uninhabited aerial vehicles (UAVs) are widely used not only in civil fields such as power inspection and environmental monitoring, but also in military applications such as reconnaissance, surveillance and confusion. The drone “bee colony” is composed of a group of small unmanned aerial vehicles that work together independently. It has excellent features such as low cost, high damage resistance, good sensing ability, strong collaboration ability and functional distribution, which can improve the efficiency of completing task. In the complex electromagnetic environment of the battlefield, it is especially important to ensure the flight safety between the formation of the drone group and the reliable communication within the formation. The advantages of wireless ultraviolet communication mainly include small background noise, strong anti-electromagnetic interference capability, all-weather non-direct view communication, low power consumption, high integration, easy to load, etc., which can meet the communication requirements in this environment.
This paper proposes an algorithm for collaborative avoidance using wireless ultraviolet light between drones in a bee colony drone formation. Through combining avoidance algorithm with the characteristics of wireless ultraviolet light coverage, a wireless ultraviolet virtual fence avoidance strategy is proposed. Considering the relationship between the enhanced vector field histogram method and its own motion state to improve the cost function and verify the effectiveness of the avoidance algorithm.The unscented Kalman filter predictor is used to predict the flight state of the adjacent drone in order to achieve safe and efficient avoidance. Through computer simulation in two prediction scenarios, the results show that the improved enhanced vector field histogram method has smooth overall motion trajectory and good avoidance effect. Compared with the original algorithm, this algorithm has no obvious jitter when it is partially avoided, the turning arc is large and there is no sharp turn. It is more suitable for the actual application and reduces the path length and time consumption.In summary, in the complex battlefield environment, the bee swarm drone can not only use airborne wireless ultraviolet equipment to achieve stable network communication,it can also use improved enhanced vector field methods based on wireless ultraviolet light to enable efficient avoidance between drones in a bee colony drone formation.
Citation: Zhao T F, Gao P, Shi H Q,An algorithm for the bee colony drone to use wireless ultraviolet for cooperative obstacle avoidance[J]., 2020, 47(3): 190505
An algorithm for the bee colony drone to use wireless ultraviolet for cooperative obstacle avoidance
Zhao Taifei1,3*, Gao Peng1,3, Shi Haiquan1,3, Li Xingshan2
1School of Automation and Information Engineering, Xi¢an University of Technology, Xi¢an, Shaanxi 710048, China;2General Design Institute of Hubei Academy of Aerospace Technology, Wuhan, Hubei 430040, China;3Shanxi Civil-Military Integration Key Laboratory of Intelligence Collaborative Networks, Xi¢an, Shaanxi 710048, China
For complex battlefield environments, it is especially important to ensure the safety of flight between uninhabited aerial vehicles (UAV) formations and reliable communication within the formation. This paper proposes an algorithm for collaborative avoidance using wireless ultraviolet light between drones in a bee colony drone formation. Combined with the above algorithm and using the characteristics of wireless ultraviolet light coverage, the avoidance strategy of ultraviolet virtual fence is designed. And by enhancing the vector field histogram method to improve the cost function of the state of motion of the drone when performing mutual avoidance. In addition, the algorithm uses the unscented Kalman filter to predict the flight status of nearby uninhabited aerial vehicles. The simulation results show that in the avoidance simulations of the two prediction scenarios, the overall motion trajectory of this algorithm is smoother than that of the enhance vector field histogram method. At the same time, there is no obvious jitter when local avoidance occurs, the total length of the avoidance path is reduced by 3.46% on average, and the total time consumption is reduced by 18.94%. This verifies that the wireless ultraviolet cooperative avoidance algorithm in a bee colony drone formation is effective.
colony drone; wireless ultraviolete; virtual fence; cooperative obstacle avoidance; trajectory prediction; enhanced vector
V279+.2;TN929.1
A
10.12086/oee.2020.190505
: Zhao T F, Gao P, Shi H Q,. An algorithm for the bee colony drone to use wireless ultraviolet for cooperative obstacle avoidance[J]., 2020,47(3): 190505
2019-08-26;
2019-09-26
國家自然科學基金資助項目(61971345,U1433110);陜西省教育廳服務地方專項計劃項目(17JF024);西安市科學計劃項目(CXY1835(4));陜西省重點產(chǎn)業(yè)鏈創(chuàng)新計劃項目(2017ZDCXL-GY-05-03);西安市碑林區(qū)科技計劃項目(GX1921)
趙太飛(1978-),男,博士,教授,主要從事網(wǎng)絡通信與自組織網(wǎng)絡技術的研究。E-mail:tfz@xaut.edu.cn
趙太飛,高鵬,史海泉,等. 蜂群無人機編隊內無線紫外光協(xié)作避讓算法[J]. 光電工程,2020,47(3): 190505
Supported by National Natural Science Foundation of China (61971345, U1433110), Shaanxi Provincial Department of Education Service Local Special Project (17JF024), Xi¢an Science Project (CXY1835(4)), Shaanxi Provincial Key Industry Chain Innovation Project (2017ZDCXL-GY-05-03), and Xi¢an Beilin District Science and Technology Plan Project (GX1921)
* E-mail: tfz@xaut.edu.cn