A fuzzy-augmented Kalman filter for IMU/GPS integration

Most of the present techniques for integrating Inertial Measurement Units (IMU) and Global Positioning Systems (GPS) utilize Kalman filtering (KF) as the integration estimation technique. KF is a recursive algorithm designed to compute corrections to a system based on external measurements. In inertial navigation, this can be accomplished by using an external navigation reference such as GPS. As long as GPS measurements are available, the KF solution of IMU/GPS integration works efficiently and provides accurate estimate of the navigation states. Nevertheless, during GPS signal outages, the functionality of KF update engine is disrupted due to the lack of GPS update measurements and therefore KF works only in prediction mode. Moreover, IMUs, particularly those integrating low-cost sensors, suffer from one serious limitation: drift rate errors rapidly accumulate with the passage of time. As a result, the corresponding state estimate will also quickly drift over time causing a dramatic degradation in the overall accuracy of the integrated system. Performance improvements of integrated IMUs, utilizing low-cost sensors, and GPS are presented in this paper. This achieved through the implantation of a new technique which augment KF and Fuzzy logic principles. In the innovation in the new technique is in its ability to generate the update measurements (positions and velocity error measurements) to the KF update engine even during GPS signal outages. This proposed technique has been tested on real MEMS inertial and GPS data collected in a land vehicle navigation test. The test results indicate that the proposed Fuzzy model can efficiently compensate for GPS updates during short GPS signal outages.