This paper proposes a new method called the generalized incremental predictive Kalman filter (GIPKF) to increase the accuracy of the integrated GPS / INS systems when the satellite signal is not available. This method is performed in the laboratory and tested and evaluated using the prepared hardware. The equations governing the inertial navigation system are nonlinear and the linearization in the extended Kalman filter causes the linearization approximation error. The uncertainties in the measurement noises and system noises also, produce errors in the estimation. In the proposed method, the model errors such as the linearization error and the weighted noise matrices errors are assumed as the model’s filter error and are estimated and compensated using the concept of predictive filtering and the application of Kalman filter. In this paper first, the complete equations of the new proposed method and the relations required to integrate the GPS/INS system are explained. Then using the results of the experiments, the proposed method is compared to the extended Kalman filter method. The results show that the presented algorithm is more efficient since, when the GPS outage is about 30 seconds, the position error is reduced by about 50% due to the new method’s ability to predict and compensate for the model error. This method significantly improves the performance of inertial navigation systems.
In this paper, the complete equations of the proposed new method and the relations required to integrate the GPS/INS system are first explained. Then, using the test results, the proposed new method is compared with extended Kalman filter method. The results show that the presented algorithm is more efficient when the receiver signals are blocked due to its ability to predict and compensate for model error. This method significantly improves the performance of the inertial navigation system and corrects its errors.