GNSS/UWB integrated positioning with robust Helmert variance component estimation.

A robust GNSS/UWB integration algorithm is proposed for the indoor-outdoor cross region to control the influences of observation outliers of both GNSS and UWB. An important attempt in this paper is to re-estimate the uncertainty of the virtual observations derived from the kinematic model predicted information, and the GNSS satellites as well as the UWB anchors observations by using variance component estimation. Due to the correlations of the derived observations from estimated state parameters at previous epoch, a bi-factor inflated covariance model is employed to reduce the influences of the predicted correlated observations. Accordingly, a signal variance inflation factor is employed for independent measurements. Semi-simulation and field tests are carried out to validate and analyze the performance of the proposed algorithms in terms of positioning accuracy. The simulation demonstrates that the proposed algorithm improves the positioning accuracy when outliers exist in the measurements and the correlated predicted information. The field test results show that the positioning accuracy by using the proposed methods is improved by about 56% in average of E, N, and U components compared to that of the least squares positioning strategy, and about 40% in the average of E, N, and U components compared to that of integrated positioning by EKF without companying the variance component estimation. [ABSTRACT FROM AUTHOR]