A Novel Algorithm for Enhancing the Navigation Accuracy of INS/GNSS Integrated Systems in Urban Cities

For the inertial navigation system (INS)/ the global navigation satellite system (GNSS) integration, the stochastic error of the inertial sensor casts a significant influence on the overall positioning performance. Since stochastic error of inertial sensors generally has a quite complex structure, the classical estimation techniques including the Allan variance (AV) method, and the power spectral density (PSD) analysis suffer from the obvious drawbacks. However, a recently proposed estimator called the generalized method of wavelet moments (GMWM) is utilized to model the stochastic error for inertial sensors. Focused on the further improvement of GMWM and the design of the stochastic error model within the integrated navigation filter, a novel algorithm for enhancing the navigation accuracy of INS/GNSS integrated systems in urban cities is proposed. We performed a practical car experiments in urban areas to analyze the navigation solution of this novel algorithm. Then comparison with the AV-based method and the classical EKF method is further discussed. The results verified that the proposed novel algorithm can successfully enhance the navigation accuracy and implement the automatic modeling of the complicated stochastic errors.