Improved GPS/IMU Loosely Coupled Integration Scheme Using Two Kalman Filter-based Cascaded Stages

Despite the fact that accelerometers and gyroscopes are used in inertial navigation systems (INS) to provide navigation information without the aid of external references, accumulated systematic errors are shown in sensor readings on long-term usage. In this work, a new approach is proposed to overcome this problem, by using extended Kalman filter (EKF)—linear Kalman filter (LKF), in a cascaded form, to couple the GPS with INS. GPS raw data are fused with noisy Euler angles coming from the inertial measurement unit (IMU) readings, in order to produce more consistent and accurate real-time navigation information. The proposed algorithm is designed to run through three software threads simultaneously. The multi-thread processing provides better use of hardware resources and applies more efficient INS/GPS loosely coupled integration scheme compared to the conventional method. Two datasets are used to verify the efficacy of the proposed approach against the existing GPS/INS coupling techniques. The first set is synthetic data generated by MATLAB that represents a static vehicle at known coordinates. The second one is a real road test data collected in Ontario, Canada. Accordingly, the root mean square error (RMSE) values for the proposed approach in ENU directions have reached 0.022, 0.034 and 0.010 m, respectively, for a static vehicle, as well as 0.493, 0.453 and 0.110 m, respectively, for a movable vehicle—which is notably competitive with other recent related work.