Gravity-Aided Transfer Alignment With Enhanced Horizontal Observability Under Large Azimuth Misalignment

Abstract

This article proposes a gravity-aided transfer alignment (TA) method within a cubature Kalman filter (CKF) framework to enhance the observability of horizontal misalignment angles under large initial azimuth errors, particularly in low-dynamic scenarios. Initially, a comprehensive nonlinear error model is established to describe the TA process, incorporating large heading misalignment, inertial sensor biases, installation errors, and structural flexural effects in a unified state-space representation. Subsequently, an acceleration-compensated gravity-related observation is derived from the master inertial navigation system (MINS) and fused with conventional velocity and attitude matching, providing complementary local sensitivity to horizontal attitude errors and accelerometer biases under weak-excitation conditions. Finally, the proposed method is validated through simulations and vehicle-mounted experiments. The results demonstrate that the gravity-aided approach achieves faster convergence and significantly improved estimation accuracy for horizontal misalignment angles while maintaining stable azimuth estimation performance under large initial heading errors.