GPS-INS outlier detection & elimination using a sliding window filter

Many applications require reliable, high precision navigation (sub-meter accuracy) while using low-cost inertial and global navigation satellite systems (GNSS). Success requires optimal state estimate while mitigating measurement outliers. Common implementations use an Extended Kalman Filter (EKF) combined with the Receiver Autonomous Integrity Monitoring (RAIM) on a single epoch. However, if the linearization point of the EKF is incorrect or if the number of residuals is too low, outlier detection decisions may be incorrect. False alarms result in good information not being incorporated. Missed detections result in incorrect information being incorporated. Either case can cause subsequent incorrect decisions in the future, possibly causing divergence, due to the state and covariance now being incorrect. This article formulates a sliding window estimator that solves the full-nonlinear Maximum A Posteriori estimate in real-time. By leveraging the resulting window of residuals, an improved fault detection and removal strategy is implemented. Sensor data is used to demonstrate the interval RAIM (iRAIM) performance improvement.