Error-State Transformation Based GNSS/MEMSSINS Integration Method for Low Maneuvering Vehicular Navigation

Due to the weak observability of the weakly maneuvering vehicles, MEMS-based GNSS/INS integrated navigation suffers from heading error divergence. This paper proposes a combined Kalman filtering method of using error-state transformation and lever-arm vector error compensation to improve system model accuracy. The error-state transformation replaces the noisy specific force terms by the steady-changing gravity term in the system dynamics model, reducing the adverse influence of the specific force noises upon the misalignment estimation. Moreover, the introduction of lever-arm vector error improves the equation structure of SINS misalignment, which contributes to the correction of attitude error with GNSS aiding. The validity of the proposed algorithm has been verified by both simulation demonstration and filed tests. The simulation results show that the strategy of error-state transformation improves the accuracy in position, velocity, and attitude. There is a significant decrease in the heading mean absolute error (MAE) from 26.9° to 5.8° with error state transformation. The field test indicates that the proposed algorithm improves the attitude MAE values from 7.6° to 2.1°.