面向复杂环境的UWB/LiDAR/IMU组合定位方法  

UWB/LiDAR/IMU combined localization method for complex

作　　者：杨秀建[1] 敖鹏 沈世全 杨义兴 皇甫尚昆 YANG Xiujian;AO Peng;SHEN Shiquan;YANG Yixing;HUANGFU Shangkun(Faculty of Transportation Engineering,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学交通工程学院,昆明650500

出　　处：《中国惯性技术学报》2024年第7期654-662,共9页Journal of Chinese Inertial Technology

基　　金：国家自然科学基金(52162046)。

摘　　要：针对室外全球卫星导航系统(GNSS)拒止并且超宽带(UWB)和LiDAR在非视距和点云特征稀疏环境下定位效果较差的问题,提出了一种面向复杂环境的UWB/LiDAR/惯性测量单元(IMU)组合定位方法。首先,利用LiDAR和UWB的互补特性设计了时变因子,用于对车辆进行重定位;然后,引入三种运动模型描述车辆的运动状态,各模型采用无迹卡尔曼滤波方法设计滤波器;最后,将重定位后的车辆定位和IMU的测量数据作为交互式多模型-无迹卡尔曼滤波算法的状态输入,解算出最终的车辆位置。实验结果表明,所提组合定位方法具有较高的定位精度,相对于UWB和LiDAR单一传感器,视距环境下的定位精度分别提升了35.1%和22.8%,非视距环境下分别提升了53.1%和27.2%;所提重定位方法相对于UWB和LiDAR单一传感器,视距环境下的定位精度分别提升了23.2%和8.7%,非视距环境下分别提升了42.9%和11.4%,体现出了较高的定位精度和对复杂环境的适应能力。Aiming at the problem that outdoor global navigation satellite system(GNSS)is rejected and ultra-wideband(UWB)and LiDAR have poor localization effect in non-line-of-sight and point cloud feature sparse environment,a UWB/LiDAR/inertial measurement unit(IMU)combined localization method for complex environment is proposed.Firstly,a time-varying factor is designed by using the complementary characteristics of LiDAR and UWB for vehicle relocation.Then,three motion model are introduced to describe the motion state of the vehicle,and the filters for each model are designed by using the unscented Kalman filtering method.Finally,the re-localized vehicle position and IMU measurement data are used as state inputs of the interactive multi-model-unscented Kalman filter algorithm to calculate the final vehicle position.The experimental results show that the proposed combined localization method presents high localization accuracy.Compared with single UWB and LiDAR sensor,the localization accuracy is improved by 35.1%and 22.8%in line-of-sight(LOS)scenario,and the corresponding improvement is 53.1%and 27.2%in non-line-of-sight(NLOS)scenario,respectively.Compared with the single UWB and LiDAR sensor,the localization accuracy of the proposed repositioning method in LOS scenario is improved by 23.2%and 8.7%,and the corresponding improvement in NLOS scenario is 42.9%and 11.4%,respectively,demonstrating high localization accuracy and adaptability to complex environments.

关 键 词：无人车辆 融合定位 超宽带 交互式多模型 无迹卡尔曼滤波 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]