A novel interactive robust filter algorithm for GNSS/SINS integrated navigation. (June 2023)

• Record Type: Journal Article
• Authors: Zhao, Bin; Zeng, Qinghua; Liu, Jianye; Gao, Chunlei; Zhao, Tianyu; Li, Rongbing
• Is Part Of: Proceedings of the Institution of Mechanical Engineers. Volume 237:Number 8(2023)
• Page Start: 1779
• Page End: 1790
• Publication Date: 2023-06
• Subjects: Integrated navigation -- strong tracking Kalman filter -- adaptive Kalman filter -- interactive robust filter -- information anomaly -- estimation accuracy; Aeronautics -- Periodicals; Astronautics -- Periodicals; Airplanes -- Design and construction -- Periodicals; Aerospace industries -- Periodicals; 629.1
• DOI: 10.1177/09544100221138133
• Languages: English
• ISSNs: 0954-4100

Abstract

To solve the problem of Kalman filter (KF) performance degradation in unmanned aerial vehicle (UAV) applications, a novel interactive robust filter algorithm for GNSS/SINS integrated navigation is proposed in this paper. The strong tracking Kalman filter (STKF) is robust to uncertain system noise but is ineffective to abnormal measurement information. Based on the same performance index function with STKF, a measurement noise covariance matrix adaptive Kalman filter algorithm (MAKF) is presented, but it is ineffective under uncertain system noise. Furthermore, the interactive robust filter algorithm based on STKF and MAKF (IF-STKF-MAKF) is proposed, given the complementary characteristics of the above two filter algorithms. The STKF and MAKF operate in parallel based on the same system model. The filter probability of each filter is updated according to the likelihood function to perform output fusion and input interaction. The simulation and experiment results demonstrate that the IF-STKF-MAKF is effective and can achieve high estimation accuracy under both system noise anomalies and measurement information anomalies. In the vehicle experiment, the position accuracy of the proposed IF-STKF-MAKF method has been improved by more than 30% compared with KF, STKF, and MAKF. This method can also be extended to land vehicles, mobile robots, etc.