Dp 373 en
Implementation of an unscented Kalman filter for INS/GPS integration and performance comparison with the standard extended Kalman filter approach
Author: Joshy Madathiparambil Jose
The objective of this thesis is to implement an unscented kalman �lter for integrating INS with GPS and to analyze and compare the results with the extended kalman �lter approach. In a loosely coupled integrated INS/GPS system, inertial measurements from an IMU (angular velocities and accelerations in body frame) are integrated by the INS to obtain a complete navigation solution and the GPS measurements are used to correct for the errors and avoid the inherent drift of the pure INS system. The standard approach is to use an extended kalman �lter in complementary form to model the errors of the INS states and use the GPS measurements to estimate corrections for these errors which are then feedback to the INS. Although the unscented kalman �lter is more computational intensive, it is supposed to outperform the extended kalman �lter and be more robust to initial errors. The main goal of this work is to analyze the di�erence in performance and robustness between both implementations. As a �rst step, a simpli�ed Attitude estimation of a stabilized platform is implemented in both, the UKF and the EKF and eventually the UKF will be implemented in a more complex realistic 3D navigation problem and compare against the current model used by Honeywell.
