A Consistent Transfer Alignment Method for Large Misalignment Angles: Modeling and Implementation Based on Rodrigues Parameters

This paper proposes a consistent and efficient transfer alignment method for master-slave inertial navigation systems (INS) with large installation error angles, leveraging Rodrigues parameters under high-dynamic conditions. To eliminate reliance on the slave INS strapdown solution, a motion decoupling strategy is introduced via integral formulations of specific force and angular velocity in the inertial-fixed frame, effectively suppressing high-frequency noise and flexural disturbances. The relative attitude is parameterized using Rodrigues vectors, and the attitude kinematics are linearized via the Cayley transformation, reducing the system state dimension from 15 to 9. An Marginalized Square-root Cubature Kalman Filter (MSCKF) is developed, which performs linear time updates directly and adopts third-order spherical-radial cubature rules for nonlinear measurement updates. This hybrid structure significantly enhances both computational efficiency and estimation accuracy. Simulation results demonstrate that, compared with the traditional fast transfer alignment method, the proposed approach achieves superior performance. Specifically, the average alignment angle error $\phi_a^x, \phi_a^y, \phi_a^z$ shows improvements of 83.8%, 15.1%, and 38.0% at integration frequencies of 1 Hz, 2 Hz, 5 Hz, and 10 Hz, respectively. The method is especially suited for high-precision applications on maneuvering platforms in complex dynamic environments.