一种基于机器学习的无人机集群协同定位方法  

A Cooperative Localization Method for Unmanned Aerial Vehicle Swarm Based on Machine Learning

作　　者：刘雅宁 卢浩 程彦汇 LIU Yaning;LU Hao;CHENG Yanhui(CEC Huada Electronic Design Co.,Ltd.,Beijing 102209,China;Shandong Yunhai Guochuang Cloud Computing Equipment Industry Innovation Co.,Ltd.,Beijing 100095,China;Beijing Smartchip Microelectronics Technology Co.,Ltd.,Beijing 102200,China)

机构地区：[1]北京中电华大电子设计有限责任公司,北京102209 [2]山东云海国创云计算装备产业创新中心有限公司,北京100095 [3]北京智芯微电子科技有限公司,北京102200

出　　处：《自动化与信息工程》2025年第2期32-37,共6页Automation & Information Engineering

摘　　要：针对无线定位技术受非视距误差等影响,导致定位误差较大的问题,提出一种基于机器学习的无人机集群协同定位方法。首先,建立无人机集群定位模型,并利用基于到达时间差(TDOA)的定位算法消除无人机与目标的时钟偏差;然后,采用支持向量机构建视距/非视距(LOS/NLOS)分类模型,并通过分析接收信号特征参数,分类LOS与NLOS信号;最后,根据分类结果,利用基于特征选择的NLOS抑制算法筛选适合参与目标定位计算的样本。仿真结果表明,该定位方法可以降低NLOS误差,提升定位算法的鲁棒性。To address the issue of significant positioning errors caused by non-line-of-sight(NLOS)propagation and other interference factors in wireless localization technologies,this paper proposes a machine learning based collaborative localization method for unmanned aerial vehicle(UAV)swarms.Firstly,a localization model for UAV clusters is established,and a time difference of arrival(TDOA)based localization algorithm is utilized to eliminate clock deviations between UAVs and the target.Then,a line-of--sight/non-line-of-sight(LOS/NLOS)classification model is constructed using a support vector machine(SVM),which distinguishes LOS and NLOS signals by analyzing characteristic parameters of received signals.Finally,based on the classification results,a feature selection based NLOS mitigation algorithm is applied to filter samples suitable for target localization calculation.Simulation results demonstrate that the proposed method effectively reduces NLOS induced errors and enhances the robustness of the localization algorithm.

关 键 词：无人机集群 协同定位 机器学习 非视距误差 支持向量机 到达时间差 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP212.9[自动化与计算机技术—计算机科学与技术]