Modified Observation Model in Tightly-Coupled INS/GPS Integration

The conventional Kalman filter (CKF) is widely used in tightly-coupled INS/GPS integrated navigation systems. The linearization accuracy of the CKF observation model is one of the decisive factors of the estimation accuracy and therefore navigation accuracy. Additionally, the conventional observation model (COM) used by the filter may be divergent, which would result into some terrible accuracies of INS/GPS integration navigation in some cases. To improve the navigation accuracy, the linearization accuracy of the COM still needs further improvement. To deal with this issue, the observation model is modified with the linearization of the range and range rate equations in this paper. Compared with COM, the modified observation model (MOM) further considers the difference between the real user position and the position calculated by SINS. To verify the advantages of this model, INS/GPS integrated navigation simulation experiments are conducted with the usage of COM and MOM respectively. According to the simulation results, the positions (velocities) calculated using COM are divergent over time while the others using MOM are convergent, which demonstrates the higher linearization accuracy of MOM.