Fuzzy Adaptive Cubature Kalman Filter for Integrated Navigation Systems

Area of Science:

• Navigation Systems Engineering
• Control Theory
• Signal Processing

Background:

• Integrated navigation systems like GPS/INS are crucial for accurate positioning.
• Kalman filter variants (EKF, UKF, CKF) are used but can degrade due to modeling errors and uncertainties.
• Selecting process noise covariance often relies on subjective experience or simulations, impacting performance.

Purpose of the Study:

• To develop an improved sensor fusion method for integrated navigation systems.
• To address the performance degradation of nonlinear filters caused by system uncertainties.
• To propose an adaptive approach for tuning the process noise covariance matrix.

Main Methods:

• A sensor fusion method combining Cubature Kalman Filter (CKF) with a Fuzzy Logic Adaptive System (FLAS).
• Application of a third-degree spherical-radial cubature rule within CKF to enhance numerical stability.
• Integration of FLAS into CKF to adaptively adjust the process noise covariance matrix using a Degree of Divergence (DOD) parameter.

Main Results:

• The proposed Fuzzy Adaptive Cubature Kalman Filter (FACKF) algorithm demonstrates significant accuracy improvements.
• FACKF shows superior performance compared to Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), and standard CKF.
• The adaptive tuning mechanism effectively mitigates performance degradation from system dynamics uncertainties.

Conclusions:

• The FACKF offers a robust and accurate solution for sensor fusion in integrated navigation.
• Adaptive adjustment of process noise covariance is key to improving nonlinear filter performance in uncertain environments.
• This method provides a more reliable alternative to conventional Kalman filtering techniques for navigation applications.