基于视觉的机场无人驱鸟车路径规划算法

王蕊; 李金洺; 史玉龙; 孙辉

doi:10.13700/j.bh.1001-5965.2022.0717

基于视觉的机场无人驱鸟车路径规划算法

doi: 10.13700/j.bh.1001-5965.2022.0717

中国民航大学电子信息与自动化学院，天津 300300

详细信息

通讯作者:
E-mail：huisun@cauc.edu.cn

中图分类号: TP751
计量
- 文章访问数: 298
- HTML全文浏览量: 103
- PDF下载量: 43
- 被引次数: 0
出版历程
- 收稿日期: 2022-08-16
- 录用日期: 2022-09-23
- 网络出版日期: 2022-12-14
- 整期出版日期: 2024-05-29

Vision-based path planning algorithm of unmanned bird-repelling vehicles in airports

College of Information Engineering and Automation，Civil Aviation University of China，Tianjin 300300，China

More Information

Corresponding author: E-mail：huisun@cauc.edu.cn

摘要

摘要:
机场飞行区存在的低空飞鸟严重威胁飞行器的起飞和降落安全，现有的驱鸟措施难以高效驱离低空飞鸟，且存在设备资源消耗高、受时空影响大等问题。为此，使用无人驱鸟车替代有人驾驶车辆进行驱鸟工作，并使用搭载固定摄像云台的无人驱鸟车对机场低空中鸟类进行实时检测，获取鸟情数据后，为无人驱鸟车路径规划提供鸟情数据基础。针对鸟类检测的问题，提出一种基于坐标注意力机制改进的YOLOv5网络，对小目标鸟类进行高效的实时检测，使网络更加精准地对鸟类进行定位；针对传统路径规划算法存在路径距离较长、拐点较多等缺陷，提出一种改进的天牛群算法，可有效缩短无人驱鸟车行驶距离，精准躲避机场内静态障碍物和动态障碍物，并快速到达指定驱鸟位置。实验结果表明：所提算法可对机场鸟类进行有效检测，为无人驱鸟车及时提供鸟情数据，利用改进的天牛群算法缩短规划路径的距离，使无人驱鸟车更加精准快速地到达指定驱鸟位置，有效减少人力资源投入，节约无人驱鸟车行进所需能源，提高驱鸟效率。
- 卷积神经网络 /
- 鸟类检测 /
- 无人车 /
- 天牛群算法 /
- 路径规划
Abstract:
Low-flying birds flying in the vicinity of the airports are a serious threat to the safety of aircraft takeoff and landing, and the existing bird-repelling measurements make it difficult to effectively repel low-flying birds for high instrument resource consumption and large spatio-temporal influence. In order to reduce the workload associated with repelling birds, this paper suggests replacing manned vehicles with unmanned vehicles. These unmanned vehicles will be outfitted with fixed cameras to enable real-time bird detection near the airport, as well as the collection and provision of bird data for the unmanned vehicles’ route planning. The method is divided into two parts: bird detection and path planning of unmanned bird-repelling vehicles. In order to enhance the accuracy of the network’s bird location, this study first addresses bird detection. Specifically, it suggests an enhanced YOLOv5 network that utilizes a coordinate attention mechanism to effectively identify small target birds in real time. Second, in view of the path planning problem of unmanned bird-repelling vehicles, the traditional path planning algorithms need to be improved in perspectives of long path distances and more inflection points. Therefore, an improved beetle swarm optimization algorithm is proposed in this paper, which can effectively shorten the marched distance of unmanned bird-repelling vehicles, accurately avoid static obstacles and dynamic obstacles in the airport, and quickly reach the designated location. The results show that the method can effectively detect airport birds, and provide timely bird data for unmanned bird-repelling vehicles. The route planning distance can be shortened by using the enhanced beetle swarm optimization technique, giving unmanned bird-repelling vehicles quick access to designated locations. It can effectively reduce human resource investments, save the unmanned bird-repelling vehicles energy, and improve the bird-repelling efficiency.
- convolutional neural networks /
- bird detections /
- unmanned ground vehicle /
- beetle swarm optimization algorithm /
- path planning

HTML全文

图 1 坐标注意力机制

Figure 1. Coordinate attention mechanism

下载: 全尺寸图片幻灯片

图 2 路径规划环境建模示意图

Figure 2. Schematic diagram of environmental modeling for path planning

下载: 全尺寸图片幻灯片

图 3 改进天牛群算法流程

Figure 3. Flow of improved beetle swarm optimization algorithm

下载: 全尺寸图片幻灯片

图 4 基于天气因素的数据增强

Figure 4. Data enhancement based on weather factors

下载: 全尺寸图片幻灯片

图 5 4种算法静态环境下路径规划仿真结果

Figure 5. Simulation results of path planning in static environment compared with four algorithms

下载: 全尺寸图片幻灯片

图 6 存在动态障碍物的路径规划仿真结果

Figure 6. Simulation results of path planning in the presence of dynamic obstacles

下载: 全尺寸图片幻灯片

表 1 改进的CSPDarkNet53主干网络结构

Table 1. Improved CSPDarknet53 backbone network architecture

模块	数量	参数量	尺寸	输出尺寸
Focus	1	3520	1×1	320×320
Conv	1	18560	3×3	160×160
C3	3	18816		160×160
CA	1	1688		160×160
Conv	1	73984	3×3	80×80
C3	9	156928		80×80
CA	1	3352		80×80
Conv	1	295424	3×3	40×40
C3	9	625152		40×40
CA	1	6680		40×40
Conv	1	1180672	3×3	20×20
SPP	1	656896	1×1, 5×5, 9×9, 13×13	20×20
C3	3	1182720		20×20

下载: 导出CSV

表 2 鸟类检测实验结果对比

Table 2. Comparison of bird detection experimental results

方法	召回率	精准率	平均精度	检测速度/(帧·s⁻¹)
YOLOv5	0.823	0.841	0.874	57.2
改进YOLOv5	0.902	0.914	0.931	55.9

下载: 导出CSV

表 3 4种算法运行100次的路径长度平均值

Table 3. Average path length of the four algorithms for 100 runs

算法	路径长度/m
A-star	86.122
PSO	89.725
IBSO	84.736
IBSO2	83.026

下载: 导出CSV

参考文献(25)

[1]	HE K, ZHANG X, REN S, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(9): 1904-1916. doi: 10.1109/TPAMI.2015.2389824
[2]	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2014: 580-587.
[3]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: Single shot multibox detector[C]//Proceedings of the European Conference on Computer Vision. Berlin: Springer, 2016: 21-37.
[4]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2016: 779-788.
[5]	REDMON J, FARHADI A. YOLO9000: Better, faster, stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2017: 7263-7271.
[6]	REDMON J, FARHADI A. YOLOv3: An incremental improvement[EB/OL]. (2018-04-08)[2022-05-25]. https://arxiv.org/abs/1804.02767.
[7]	BOCHKOVSKIY A, WANG C Y, LIAO H Y. YOLOv4: Optimal speed and accuracy of object detection[EB/OL]. (2020-04-23)[2022-05-25]. https://arxiv.org/abs/2004.10934.
[8]	HOU Q, ZHOU D, FENG J. Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2021: 13713-13722.
[9]	赵江, 张岩, 马泽文, 等. 对AGV路径规划A星算法的改进与验证[J]. 计算机工程与应用, 2018, 54(21): 217-223. ZHAO J, ZHANG Y, MA Z W, et al. Improvement and verification of A-star algorithm for AGV path planning[J]. Computer Engineering and Applications, 2018, 54(21): 217-223(in Chinese).
[10]	TANG G, TANG C Q, CLARAMUNT C, et al. A-star algorithm: An improved A-star algorithm for AGV path planning in a port environment[J]. IEEE Access, 2021, 9: 59196-59210 . doi: 10.1109/ACCESS.2021.3070054
[11]	OROZCO-ROSAS U, PICOS K, MONTIEL O. Hybrid path planning algorithm based on membrane pseudo-bacterial potential field for autonomous mobile robots[J]. IEEE Access, 2019, 7: 156787-156803. doi: 10.1109/ACCESS.2019.2949835
[12]	张万绪, 张向兰, 李莹. 基于改进粒子群算法的智能机器人路径规划[J]. 计算机应用, 2014, 34(2): 510-513. ZHANG W X, ZHANG X L, LI Y. Path planning for intelligent robots based on improved particle swarm optimization algorithm[J]. Journal of Computer Applications, 2014, 34(2): 510-513(in Chinese).
[13]	柳长安, 鄢小虎, 刘春阳, 等. 基于改进蚁群算法的移动机器人动态路径规划方法[J]. 电子学报, 2011, 39(5): 1220-1224. LIU C A, YAN X H, LIU C Y, et al. Dynamic path planning for mobile robot based on improved ant colony optimization algorithm[J]. Acta Electronica Sinica, 2011, 39(5): 1220-1224(in Chinese).
[14]	VIEIRA R, ARGENTO E, REVOREDO T. Trajectory planning for car-like robots through curve parametrization and genetic algorithm optimization with applications to autonomous parking[J]. IEEE Latin America Transactions, 2022, 20(2): 309-316. doi: 10.1109/TLA.2022.9661471
[15]	杨俊成, 李淑霞, 蔡增玉. 路径规划算法的研究与发展[J]. 控制工程, 2017, 24(7): 1473-1480. YANG J C, LI S X, CAI Z Y. Research and development of path planning algorithm[J]. Control Engineering of China, 2017, 24(7): 1473-1480(in Chinese).
[16]	魏彤, 龙琛. 基于改进遗传算法的移动机器人路径规划[J]. 北京航空航天大学学报, 2020, 46(4): 703-711. WEI T, LONG C. Path planning for mobile robot based on improved genetic algorithm[J]. Journal of Beijing University of Aeronautics and Astronautics, 2020, 46(4): 703-711(in Chinese).
[17]	段安娜, 周锐, 邸斌. 考虑先验信息的多机器人重点区域协同覆盖[J]. 北京航空航天大学学报, 2023, 49(6): 1479-1486. DUAN A N, ZHOU R, DI B. Multi-robot cooperative coverage of key regions considering priorinformation[J]. Journal of Beijing University of Aeronautics and Astronautics, 2023, 49(6): 1479-1486(in Chinese).
[18]	赵萌, 路辉, 王诗琪, 等. 多种群合作学习的多模态多目标路径规划算法[J]. 北京航空航天大学学报, 2023, 49(3): 606-616. ZHAO M, LU H, WANG S Q, et al. A multimodal multi-objective path planning algorithm based on multi-swarm cooperative learning[J]. Journal of Beijing University of Aeronautics and Astronautics, 2023, 49(3): 606-616(in Chinese).
[19]	孙波, 陈卫东, 席裕庚. 基于粒子群优化算法的移动机器人全局路径规划[J]. 控制与决策, 2005, 20(9): 1052-1055. SUN B, CHEN W D, XI Y G. Particle swarm optimization based global path planning for mobile robots[J]. Control and Decision, 2005, 20(9): 1052-1055(in Chinese).
[20]	TONG Y F, ZHONG M, LI J W, et al. Research on intelligent welding robot path optimization based on GA and PSO algorithms[J]. IEEE Access, 2018, 6: 65397-65404. doi: 10.1109/ACCESS.2018.2878615
[21]	赵辉, 郝梦雅, 王红君, 等. 基于改进A算法与天牛须搜索算法的农业机器人路径规划方法[J]. 科学技术与工程, 2019, 19(31): 185-190. ZHAO H, HAO M Y, WANG H J, et al. Path planning method of agricultural robot based on improved A algorithm and beetle antennae search algorithm[J]. Science Technology and Engineering, 2019, 19(31): 185-190(in Chinese).
[22]	HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE Press, 2018: 7132-7141.
[23]	WOO S, PARK J, LEE J Y, et al. CBAM: Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision. Berlin: Springer, 2018: 3-19.
[24]	单梁, 强浩, 李军, 等. 基于Tent映射的混沌优化算法[J]. 控制与决策, 2005, 20(2): 179-182. SHAN L, QIANG H, LI J, et al. Chaos optimization algorithm based Tent map[J]. Control and Decision, 2005, 20(2): 179-182(in Chinese).
[25]	WANG R, LI J M. A new improved beetle swarm optimization algorithm[C]//Proceedings of the Chinese Intelligent Systems Conference. Berlin: Springer, 2021, 805: 663-671.