Improving Loosely Coupled GNSS/IMU Fusion Performance with Pseudorange Error Prediction in Urban Areas

Abstract

AbstractPositioning performance of Global Navigation Satellite System (GNSS) will be severely degraded by multipath in urban environments. Integration GNSS with Inertial Measurement Unit (IMU) is a widely used method to improve the positioning performance. Nevertheless, the performance of the fusion results can be restricted by setting of the parameters in the filtering algorithm. In particular, the proper setting of the GNSS measurement noise covariance is usually a critical issue in the rapidly changing urban environment. In order to achieve an improved navigation performance in urban areas, we have proposed a new GNSS update strategy in loosely coupled GNSS/IMU fusion scheme based on the average of the predicted pseudorange errors and the value of Position Dilution of Precision (PDOP). The ensemble bagged tree algorithm is used to fit pseudorange errors by considering multiple variables including coordinate information, satellite signal strength and elevation angle. Furthermore, base the average of predicted pseudorange error, the measurement noise adjustment scheme is proposed for the fusion algorithm. The proposed algorithm can provide improvements of 28.15% and 43.10% compared to the traditional GNSS/IMU integrated algorithm in the horizontal and 3D positioning results.KeywordsAdaptive Kalman filterPseudorange error predictionEnsemble bagged tree