A Study of Unscented Kalman Filter Performance in the Ultralight Coupled Integration System

Abstract

This paper aims to the enhancement of the performance in the navigation filter of the GNSS/INS ultralight coupled (UTC) integration system. The basic structure of the UTC and the state-space equations are provided in this paper. Owing to its strong nonlinearity, a kind of nonlinear Kalman filter based on the unscented transformation is introduced to the system, which can lead to higher accuracy and stronger robustness than traditional EKF. Furthermore, when applying into a realistic nonlinear system without detailed information about the model and noise, an advanced UKF is proposed in this paper to deal with the potential increasing state-error and divergence of the system. Finally, Simulations are made to prove the availability of these theories above.