基于超宽带的多传感器室内组合定位算法研究  

Research on Multi-sensor Indoor Integrated Positioning Algorithm Based on UWB

作　　者：沈开淦 侯志宽 陈依玲 陈帅[1] 向峥嵘[1] SHEN Kaigan;HOU Zhikuan;CHEN Yiling;CHEN Shuai;XIANG Zhengrong(Nanjing University of Science and Technology,Nanjing 210094)

机构地区：[1]南京理工大学自动化学院,南京210094

出　　处：《导航与控制》2025年第1期71-82,共12页Navigation and Control

摘　　要：在GNSS信号拒止的复杂室内环境中,超宽带(Ultra Wideband,UWB)技术以其高精度和高稳定性等优势备受关注。针对室内复杂环境下无人车的定位问题,以惯性导航系统(Inertial Navigation System,INS)和UWB为基础,提出了基于自适应矢量分配联邦卡尔曼滤波(Adaptive Federated Kalman Filter Algorithm for Line-of-sight Reconstruction,Re-Los-AFKF)的INS/UWB/高度计/磁罗盘组合导航算法。算法建立了各子滤波器的数学模型,使用了基于矢量的自适应信息分配因子算法。设计了INS/UWB紧组合子滤波器,利用INS短时精度高的特点预测无人车的位置参数实现了UWB测距信息中的非视距(Non Line-of-sight,NLOS)误差检测,使用相邻时刻的数据对NLOS信息进行视距重构,并对重构后的距离信息进行过补偿识别和过补偿矫正。实验结果表明,在视距(Line-of-sight,LOS)环境中,Re-Los-AFKF的定位精度与基于双边双向测距(High Double Sided Two-way Ranging,HDS-TWR)的三边定位算法和基于平滑处理的联邦卡尔曼滤波(Movmean-federated Kalamn Filter,MFKF)算法的定位精度相当。在复杂的NLOS环境中,相较于HDS-TWR和MFKF,Re-Los-AFKF在水平位置上的误差均值分别减小了46.24%和29.35%,在水平位置上的误差RMS分别减小了41.40%和29.18%。研究表明,该算法不仅在LOS环境中能够进行稳定定位,在NLOS环境中也具有良好的定位性能,具有一定的适应性、鲁棒性和工程应用性。In complex indoor environments where GNSS signals are rejected,ultra wideband(UWB)technology has attracted much attention for its advantages such as having high accuracy and high stability.Aiming at the problem of unmanned vehicle localization in indoor complex environments,a combined INS/UWB/altimeter/magnetic compass navigation algorithm based on adaptive federated Kalman filter algorithm for line-of-sight reconstruction(Re-Los-AFKF)is proposed based on inertial navigation system and UWB.The algorithm establishes the mathematical model of each sub-filter and uses a vector-based adaptive information allocation factor algorithm.The INS/UWB tight combination sub-filter is designed to predict the position parameters of the unmanned vehicle by using the characteristics of INS with high short-time accuracy to realize the non line-of-sight(NLOS)error detection in the UWB ranging information,use the data of adjacent moments to reconstruct the NLOS information for the line-of-sight(LOS)range,and carry out the over-compensation identification and the over-compensation correction for the reconstructed distance information.The experimental results show that the localization accuracy of the Re-Los-AFKF algorithm is comparable to that of high double sided two-way ranging(HDS-TWR)and movmean-federated Kalamn filter(MFKF)in LOS environment.In the complex NLOS environment,Re-Los-AFKF reduces the mean value of error in horizontal position by 46.24%and 29.35%,and reduces the RMS of error in horizontal position by 41.40%and 29.18%compared to HDS-TWR and MFKF,respectively.In summary,it is shown that the algorithm is not only capable of stable localization in LOS environment,but also has good localization performance in NLOS environment,which has certain adaptability,robustness and engineering applications.

关 键 词：无人车 室内环境 非视距误差 联邦卡尔曼滤波 超宽带 紧组合算法 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]