The observability analysis and SPKF for the in-motion alignment of the loosely-integrated GPS/INS system

To achieve higher navigation accuracy it is necessary to align an inertial navigation system (INS) before it commences operation. With the aid of external information, in-motion alignment can improve the accuracy of the inertial navigation solution. This paper studies the in-motion alignment for the integration of Global Positioning System (GPS) and INS technologies, with a focus on the observability analysis for in-motion alignment of GPS/INS integrated systems during maneuvres based on a perturbation model. Analysis is performed on four types of maneuvres: uniform speed, uniform acceleration, U-turn and figure-eight. The observability analysis shows that yaw-change or acceleration-change enhances the observability of misalignment angles. In order to avoid the need to linearise the nonlinear state equations and to reduce the modelling errors, the full state space approach for in-motion alignment of integrated GPS/INS is described. To improve the accuracy of alignment, the Sigma Points Kalman Fiter (SPKF) is applied to in-motion alignment. The state vector includes position, velocity, attitude quaternion, and sensor biases. GPS position and velocity are used as input measurements. Test results from a vehicle experiment show that in the case of static and straight movement, there is not much difference between a SPKF and an Extended Kalman Filter (EKF). However, for fast and extreme maneuvres, the SPKF can improve the alignment accuracy by one order of magnitude for both position and velocity in comparison to the EKF. The results suggest that the SPKF is preferred for accurate GPS/INS in-motion alignment.