Enhanced robust state estimator with adaptive parameter tuning for high-accuracy in-flight alignment of low-cost SINS/GPS

High-accuracy in-flight alignment (IFA) for airborne SINS/GPS is critical but often degraded by MEMS sensor biases and GPS measurement outliers. This paper proposes an enhanced robust state estimator with adaptive parameter tuning to address these challenges. The method introduces three key innovations within an error-state extended Kalman filter framework. First, both the full-integral velocity integration formula (VIF) and position integration formula (PIF) reference vectors are incorporated as augmented states, enabling simultaneous observation of VIF and PIF measurements to improve alignment accuracy. Second, a residual covariance redistribution method based on the normalized error vector and Huber's robust function enhances robustness against GPS outliers. Third, an adaptive parameter tuning method for the Huber threshold leverages online statistical characteristics to effectively mitigate both spike-type and, crucially, slow-drift GPS outliers—a key advantage over existing methods that typically address only one outlier type. Validated using real flight data collected from a low-cost MEMS IMU/GPS with artificially injected outliers via Monte Carlo simulations, the proposed estimator achieves the lowest root mean square error across all attitude angles. Specifically, it attains 0.28° yaw accuracy under normal conditions, 0.43° under spike outliers, and 1.65° under slow-drift outliers—outperforming classical IFA approaches, state-of-the-art sliding-window-integration-based non-augmented ESEKF methods, and robust methods that only use the VIF reference vector as augmented states. These results establish it as a robust solution for IFA in GPS-challenged environments.