An adaptive dual Kalman filtering algorithm for Locata/GPS/INS integrated navigation
Zebo Zhou,Ling Yang,Yong Li
DOI: https://doi.org/10.1007/978-3-642-37407-4_50
2013-01-01
Abstract:In modern navigation applications, Inertial Navigation System (INS) is attractive for integrating with Global Positioning System (GPS). Although traditional GPS/INS integrated systems can bridge the GPS gaps, the navigation capability is strongly dependent on the performances of standalone INS. As an important complementary, a new terrestrial, Radio-Frequency (RF) based, distance measurement technology "Locata" can provide continuously time-synchronised ranging signals even in many GPS challenged environments e.g. poor satellite geometry, signal blockage in suburban, tunnels, high rise buildings canyon. This paper investigates the integration of Locata, GPS, and INS with a focus on the loosely-coupled triple integration algorithm. Firstly, the Conventional Kalman Filtering (CKF) based triple integration of Locata/GPS/INS architecture is described and briefly discussed. Secondly, to overcome the pitfalls of conventional Locata/GPS/INS integration algorithm, an Adaptive Dual Kalman Filtering (ADKF) algorithm is proposed and developed in three stages: (1) To enhance the reliability of position and velocity (PV) quantities generated from Locata/GPS integrated sensors, the 1st KF is additionally constructed to reliably estimate the PV solution before fusing INS sensor. (2) Combining the 15-state INS error model and the measurements which are the differences between PV solution from the 1st KF and INS sensor, the 2nd KF is subsequently employed to correct the INS navigation errors. (3) The final integration solution is reversely used as the feedback for precisely estimating the stochastic model (i.e. variance of dynamic and observation model noise) of the 1st KF. Finally, the real flight experiment is carried out to demonstrate the efficiency and validity of Locata/GPS/INS integration algorithms. The results show that: (1) Conventional GPS/INS integration performs well but its accuracy dramatically decreases when GPS signals are unavailable for a short period. (2) Augmented by Locata, GPS/INS produces tolerable results in whole experiment even without aiding of GPS. (3) By additionally operation of the 1st KF and adaptively estimating its stochastic model with feedback of integration solution, ADKF achieves more accurate and reliable position, velocity and attitude (PVA) solution than CKF.