UWB/MIMU自适应动量GD-PF室内协同定位  

UWB/MIMU adaptive momentum GD-PF indoor cooperative positioning

作　　者：李鹏[1] 凌智琛 荣冬成 向宇翔 LI Peng;LING Zhichen;RONG Dongcheng;XIANG Yuxiang(School of automation and electronic information,Xiangtan University,Xiangtan,Hunan 411100,China)

机构地区：[1]湘潭大学自动化与电子信息学院,湖南湘潭411100

出　　处：《导航定位学报》2022年第6期122-128,共7页Journal of Navigation and Positioning

基　　金：国家重点研发计划项目(2020YFA0713501);湖南省自然科学基金项目(2021JJ50126);湖南省教育厅重点项目(21A0083)。

摘　　要：针对超宽带(UWB)室内定位时容易受到非视距误差影响,导致定位精度大大下降,甚至无法准确定位的问题,提出了一种UWB/MIMU自适应动量梯度下降-粒子滤波(GD-PF)室内协同定位方法。首先在梯度下降算法中引入指数加权平均和变步长策略,加快解算速度;然后采用粒子滤波对UWB解算数据进行优化,减小粗差对定位精度的影响;最后引入自适应函数调整扩展卡尔曼滤波(EKF)增益,对UWB和微型惯性测量单元(MIMU)的定位数据进行联合滤波。UWB/MIMU协同定位实验结果表明,自适应动量GD-PF协同定位算法与传统定位算法相比,能有效消除非视距误差的干扰,提高室内定位精度和鲁棒性。To address the problem that ultra wide band(UWB) indoor positioning is easily affected by non line of sight(NLOS) errors, which leads to a significant decrease in positioning accuracy or even inability to locate accurately. In this paper,an UWB/MIMU adaptive momentum gradient descent-particle filter(GD-PF) indoor co-localization method is proposed. Firstly,exponential weighted average and variable step size strategies are introduced in the gradient descent(GD) algorithm to speed up the gradient solving speed. Then the particle filtering is used to optimize the UWB solved data to reduce the gross errors on the localization accuracy. Finally, an adaptive function is introduced to adjust the extended Kalman filter(EKF) gain to jointly filter the localization data of UWB and miniature inertial measurement unit(MIMU). The experimental results of UWB/MIMU colocalization show that the adaptive momentum GD-PF co-localization algorithm can effectively eliminate the interference of NLOS errors and improve the indoor positioning accuracy and robustness compared with the traditional positioning algorithm.

关 键 词：超宽带 微型惯性测量单元 梯度下降算法 粒子滤波 自适应扩展卡尔曼滤波 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TN925[自动化与计算机技术—控制科学与工程]