A nonlinear observer for integration of GPS and inertial navigation systems

Abstract

GPS and INS have complementary properties and they are therefore well suited for integration. The integrated solution offers better long term accuracy than a stand-alone INS, and better integrity, availability and continuity than a standalone GPS receiver, making it suitable for demanding applications. The integrated filter is nonlinear both in state and measurements, and the extended Kalman-filter has been used with good results. But the extended Kalman-filter has not been proven globally stable, and it is also computationally intensive, especially within a direct integration architecture. In this work exponential stability of the origin of the combined attitude and velocity error systems is proven along with robust stability in the presence of noise and unmodelled dynamics.