An improved real-time adaptive Kalman filter for low-cost integrated GPS/INS navigation

Recent years, we have witnessed a significant progress in integrated Global Positioning System (GPS)/ Inertial Navigation System (INS) navigation systems. In such systems, Kalman filter has been playing a key role in fusing data from multiple sensors for better accuracy. Despite the success, there is still a strong need for cost-efficient solutions with acceptable precision. One of the essential challenges for such demand is that, conventional Kalman filters tend to diverge when constant noise covariance matrices no longer match the error estimation of low-cost devices. To deliver a higher level of accuracy and stability, adaptive Kalman filters have attracted considerable research effort. Based on the analysis of two recent adaptive Kalman filters, we propose an improved real-time adaptive algorithm with fuzzy logic adaptive tuning to achieve high accuracy on cost-efficient GPS/INS devices. A GPS data validation method is also introduced to reject corrupted GPS data. Real road experimental results prove that the proposed improved adaptive algorithm offers higher overall system performance.