Multi-IMUs and GNSS integrated positioning based on second-order autoregressive model

Abstract

Abstract The positioning accuracy of the low-cost inertial measurement unit (IMU) and global navigation satellite system (GNSS) integrated system degrades rapidly during the GNSS blockage, due to complex error terms contained in the IMU measurements. This study proposes a data fusion strategy by integrating measurements of multiple low-cost IMUs, aiming to denoise the high-frequency noise on individual IMU measurements. Moreover, a systematic error compensation strategy based on the second-order auto-regressive (AR2) model is proposed to reduce the impacts of the correlated error of the low-cost IMUs measurements on positioning during the GNSS blockage. The 3 h static IMUs measurement data are collected to analyze the different measurement fusion methods and measurement error characteristics. The Allan variance and power spectral density results indicate that the random errors of the fused IMU measurements are reduced by 69.57% compared to those of the individual IMU measurements. The frequency characteristic of the fused measurements is similar to that of the 6-level wavelet denoising technique with the db4 strategy, the high-frequency noise is reduced compared to that of individual IMU measurements. GNSS/IMUs loosely coupled architecture is employed to evaluate the positioning performance during the GNSS outage. Random walk (RW), first-order Gaussian Markov (GM1), and AR2 models are employed to describe the error characteristic of the IMU stochastic error. The positioning accuracy of the fused IMUs measurements by averaging with AR2 error compensating is improved by 59.88%, 59.97%, and 45.57% compared to those of RW, GM1, and AR2 with individual IMU measurements. The dynamic test results indicate that the fused IMU measurements with the AR2 error compensation decrease the error variance and improve the positioning accuracy during the GNSS outage.