A suboptimal Kalman filter with fading factors for DGPS/MEMS-IMU/magnetic compass integrated navigation

In this paper, a new approach that performs DGPS/MEMS-IMU/magnetic compass integrated navigation based on a suboptimal Kalman filter is proposed. The purpose is bounding MEMS-IMU errors with non-drift DGPS and magnetic compass through a robust filter as long as DGPS locked information is available. Because the difficulties of establishing an accurate sensor error model for MEMS-IMU and the limitation of computer burden, the suboptimal Kalman filter with fading factors is intended for improving the robustness to the model uncertainties and assure the convergence of state estimation. It is always on-line when GPS locks so that it does not need to renew initial parameters every other interval to recalibrate MEMS-IMU. The performance of this approach is tested through computer simulations.