Adaptive square-root CKF with application to DR/LBL integrated heading estimation for HOV

Dead Reckoning (DR) and Long Base Line (LBL) are a modern method in navigation of Human Occupied Vehicles (HOV). However, the accuracy of DR system would degrade sharply, and due to the obvious error drifts of each unit involved in DR. LBL has the disadvantage of low update frequency. To improve the heading estimation of DR/LBL, this paper proposes an innovative method which could adjust state error variance matrix Q in real time dynamically. Square-root Cubature Kalman filter (SR-CKF) is used to simulate the convergence of the dynamic model of DR. And, Sage-Husa maximum a posterior (MAP) is employed in filtering progress. The simulation results of the adaptive SR-CKF and CKF are compared, which show that the method proposed in this paper can obtain a fairly accurate heading estimation.