全球导航卫星系统应用
计算机科学
卫星系统
职位(财务)
协方差矩阵
卫星导航
空中航行
协方差
实时计算
惯性导航系统
卡尔曼滤波器
定位系统
格洛纳斯
全球定位系统
控制理论(社会学)
先验与后验
导航系统
卫星
模拟
全球导航卫星系统增强
作者
Mowen Li,Adrià Rovira García,Feng Zhu,Tianhe Xu,Wenfeng Nie
标识
DOI:10.1109/jiot.2026.3727122
摘要
Tightly coupled (TC) and loosely coupled (LC) models are widely used for the integrated positioning of global navigation satellite system (GNSS) and inertial navigation system (INS). The LC model integrates the positioning solutions produced by GNSS processing and INS mechanization, resulting in the degradation of positioning accuracy when GNSS is reinitialized. The TC model integrates GNSS observations with the INS mechanization solution in a single filter, but errors of one system will propagate to the other through the correlation between the state parameters. Built upon the LC model, the Semi-TC (STC) model provides the INS mechanization position as a priori information for GNSS positioning. The current STC model can be classified into two forms depending on how the covariance matrix of the mechanization position is used in GNSS positioning, i.e., the STC-V model employs the positioning covariance matrix produced by INS mechanization, and the STC-C model uses a constant diagonal covariance matrix. Thus, the STC-V model can provide GNSS with full mechanization position information, but leads to the reuse of the mechanization position and overconstraint of positioning parameters in the integrated filter. In contrast, the STC-C model cannot rigorously constrain the GNSS positioning. Therefore, we propose a mutually coupled (MC) model featuring reciprocal constraints between INS and GNSS while avoiding the reuse of the INS mechanization position by rebuilding the integrated observation equation. In order to compare the performance of the proposed MC model with the other models, two urban vehicular GNSS/INS tests with no interruptions or faults are conducted. Furthermore, GNSS interruptions (GNSS-INTERs), GNSS observation errors (GNSS-ERRORs), and INS faults are simulated based on the original data of the two tests, which allows us to assess the performance of the MC model under different system abnormalities. The results show that the MC model performs as well as the LC, TC, and STC-C models with original nominal data. During GNSS-INTERs, the MC model can benefit the convergence of positioning errors after GNSS-INTER as what the TC model does. In the last two simulations, the MC model is capable of restraining more positioning offsets caused by GNSS errors and especially INS errors compared to the other models.
科研通智能强力驱动
Strongly Powered by AbleSci AI