FUZZY KALMAN FILTER FOR IMPROVED GPS AND GLONASS POSITIONING ACCURACY

Abstract

To enhance the GNSS/INS fusion system's performance—which is diminished in crowded urban settings owing to satellite signal cutoff and attenuation and incorrect modeling—this research proposes a fuzzy-innovation based adaptive extended Kalman filter (FI-AKF). Data for sensor fusion comes from on-board diagnostics (OBD), real-time kinematic (RTK), and micro-electro-mechanical system based inertial measurement unit (MEMS-IMU). According to the position dilution of precision (PDOP), the number of receivable satellites, and the invention of the extended Kalman filter (EKF), the measurement covariance matrix of the RTK is adaptively updated using the fuzzy logic system. Furthermore, the condition of the vehicle is described as a halt, a straight run, a left/right turn, and similar operations. The predicted heading is adjusted based on the driver's condition in order to lower the Kalman filter's heading estimation inaccuracy. In addition, the driving condition determines how the measurement covariance matrices of the IMU and OBD are applied adaptively, taking into account the properties of each sensor. A computer simulation is run to examine how well the proposed FI-AKF locating system works in a congested metropolitan setting. We evaluate the suggested FI-AKF by comparing its results with those of two popular extended Kalman filters: the current one and the one based on adaptive innovation. Furthermore, we pit our results against those of a commercial positioning system in an outdoor trial. Consistent with previous experiments, the suggested FI-AKF system outperforms the comparator positioning systems in congested metropolitan areas.