Experimental results of an adaptive fuzzy network Kalman filtering integration for low cost navigation applications

The performance of Kalman filter is highly dependent on the availability of two basic requirements: (1) accurate dynamic modeling and (2) proper measurements that fit this model. The absence of either of those two requirements will degrade the Kalman performance over time particularly during the absence of the reference signal frequently used to update the estimated Kalman states. To overcome such problem, a new design model, namely fuzzy-Kalman, integrating fuzzy logic systems and adaptive Kalman filtering for the integration of IMU and GPS is developed in this paper. The developed model was tested using an integrated GPS/IMU for land-vehicle navigation applications. The results indicated that, unlike traditional Kalman, the proposed fuzzy-Kalman model could efficiently bridge short-time outages of reference signal.