Detection and Correction of Doppler Outliers in Kalman Filter-based Positioning

Abstract

In this paper, we propose methods of detecting Doppler outliers which cause positioning errors at Doppler-aided GNSS (Global Navigation Satellite System) positioning, and correcting the errors. We apply the existing detection method based on the innovation process in Kalman filtering to Doppler outlier problems, and we propose a novel detection method based on the measurements by the difference between C/A code pseudoranges and Doppler shift range-rates. Both methods are based on chi-squared tests. We apply two correction methods which are Doppler bias exclusion, or the estimation for detected anomalies. The efficient detection of anomalous observables can be developed to RAIM (Receiver Autonomous Integrity Monitoring), and useful to achieve higher accuracy positioning for increasing satellite signals by multi-frequencies and multi-GNSSs. Doppler shift observables are utilized on a priority basis even in urban areas because of immunity to cycleslip and continuous availability, however unexpected Doppler outliers prone to cause positioning errors. The experimental results of positioning by using real receiver data show the feasibility of the proposed detection and correction methods.

Real-Time Visual Odometry Covariance Estimation for Unmanned Air Vehicle Navigation

“…This approach, using all samples under 6σ, implies that there remain delta-pose measurements with relatively large errors (i.e., greater than 3σ) in the sample distribution. However, it is assumed that a filter using these measurements will perform chi-squared (χ 2 ) or similar rejection for measurements with such large errors [22]. Otherwise, the selection of s would need to be exceedingly conservative, resulting in an undesirable distribution similar to Fig.…”