摘要
针对城市环境中无人机定位服务易受多径效应与全球导航卫星系统(GNSS)信号中断影响的问题,提出一种新颖的联合融合架构以提升其鲁棒性。该方法集成GNSS、惯性测量单元(IMU)、单目相机与气压计数据,采用扩展卡尔曼滤波器(EKF)作为局部滤波器,并创新性地以门控循环单元(GRU)神经网络块作为主滤波器进行信息融合。通过AirSim虚拟环境与Spirent GSS7000硬件分别收集视觉、气压计数据及GNSS/IMU数据,并在不同光照天气条件下进行测试,尤其关注多径与GNSS中断场景。结果表明:相较于传统扩展卡尔曼滤波(EKF)联合架构,新架构在多种退化条件下均表现出更优的弹性与鲁棒性,水平定位误差即使在灰尘天场景下也保持了0.55 m的精度,保持了良好的性能水平。
To address the problem that unmanned aerial vehicle(UAV)positioning services in urban environments are susceptible to multipath effects and global navigation satellite system(GNSS)signal interruptions,a novel joint fusion architecture is proposed to enhance its robustness.This method integrates GNSS,inertial measurement unit(IMU),monocular camera and barometer data,employing an extended Kalman filter(EKF)as a local filter and innovatively using a gated recurrent unit(GRU)neural network block as the main filter for information fusion.Visual data,barometer data and GNSS/IMU data are collected using the AirSim virtual environment and Spirent GSS7000 hardware,respectively,and tests are conducted under various lighting and weather conditions,with particular attention focused on multipath and GNSS interruption scenarios.Experimental results show that compared with the traditional EKF joint architecture,the new architecture exhibits better resilience and robustness under various degradation conditions,and the horizontal positioning error maintains an accuracy of 0.55 m even in dusty weather scenarios,maintaining a good performance level.
作者
张文娟
胡海州
杨聪敏
ZHANG Wenjuan;HU Haizhou;YANG Congmin(School of Mechanical and Electrical Engineering,Henan Vocational University of Science and Technology,Zhoukou 466000,China;School of Electrical and Information Engineering,Zhengzhou University,Zhengzhou 450001,China;College of Information and Management Science,Henan Agricultural University,Zhengzhou 450003,China)
出处
《传感器与微系统》
北大核心
2026年第4期118-123,共6页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61903341)。
关键词
无人机
导航
多传感器
联合融合架构
UAV
navigation
multi-sensor
joint fusion architecture
