为了解决传统雷达在低空小型无人机探测中探测精度受限等不足,提出一种基于深度学习的低空小型无人机探测系统。系统采用三层架构设计：感知层利用摄像头采集实时图像数据;硬件层基于边缘计算技术,使用Jetson Nano作为硬件平台,结合决策层中的目标检测算法,实现了高效且精准的预警功能。在算法选择上,通过实验对比分析了主流目标检测算法,选择最优算法YOLOv8,并对其进行优化。实验结果表明,优化后的算法各项指标均有显著提升,验证了该方案的有效性。
Abstract
To address the limitations of traditional radar in detecting low-altitude small unmanned aerial vehicles (UAVs), a deep learning-based detection system for low-altitude small UAVs is proposed. The system is designed with a three-tier architecture: the perception layer uses cameras to capture real-time image data; the hardware layer, based on edge computing technology, employs the Jetson Nano as the hardware platform; and the decision layer integrates object detection algorithms to enable efficient and accurate early warning functionality. Through experimental comparison and analysis of mainstream object detection algorithms, the optimal algorithm, YOLOv8, is selected and optimized. Experimental results show significant improvements in various performance metrics of the optimized algorithm, validating the effectiveness of the proposed solution.
关键词
无人机 /
深度学习 /
边缘计算 /
Jetson Nano /
YOLOv8
Key words
UAV /
deep learning /
edge computing /
Jetson Nano /
YOLOv8
中图分类号：
TP242
TP391.41
U463.6
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 汤新民,顾俊伟,刘冰,等.低空监视技术及其发展趋势综述[J].南京航空航天大学学报,2024, 56(6):973-993.
[2] 邱小剑,骆博雅,付珍,等.国内外反无人机技术发展综述[J].战术导弹技术,2024(5):63-73+98.
[3] 张霞峰,柳畅,单业奇,等.基于视觉传感器的无人驾驶机器人控制系统设计与研究[J].机电工程技术,2024,53(12):146-148+186.
[4] Jeong E,Kim J,Ha S,et al.TensorRT-Based Framework and Optimization Methodology for Deep Learning Inference on Jetson Boards[J].ACM Transactions on Embedded Computing Systems,2022,21(5):51-76.
[5] 张光钱,周广利,黄飞,等.面向3D目标检测的多模态生成式图像数据增强的研究[J].重庆理工大学学报(自然科学),2024,38(10):13-20.
[6] 曲坤,王震龙,刘志锋.基于多表征学习的交叉熵集成图像分类方法[J].计算机工程,2024,50(10):322-333.
[7] 孙仕棚,兰时勇.考虑注意力的无锚框孪生网络目标跟踪算法[J].计算机应用与软件,2024, 41(12):268-274.
[8] 黄海新,徐成龙.大语言模型的剪枝算法综述[J].通信与信息技术,2025(1):95-99.
[9] Jani,M.,Alhassan,et al.Model Compression Methods for YOLOv5:A Review[J].arXiv preprint.2023, arXiv:2307. 11904.
[10] Betti A.YOLO-S:A Lightweight and Accurate YOLO-like Network for Small Target Detection in Aerial Imagery[J].Sensors,23(4):1865.
基金
阜阳职业技术学院自然科学研究项目“基于深度学习的低空小型无人机探测系统研究”,项目编号：2024KYXM03; 安徽省高校自然科学研究重点项目“基于支持边缘计算的智能消防物联网网关研究”,项目编号：2023AH052417
