Robust Cubature Kalman Filter for Moving-Target Tracking with Missing Measurements

Area of Science:

• Engineering
• Control Systems
• Signal Processing

Background:

• Missing measurements pose significant challenges in nonlinear system state estimation.
• Conventional filters like the conventional cubature Kalman filter (CKF) can be limited by unknown system models and noise statistics.

Purpose of the Study:

• To propose a robust cubature Kalman filter (RCKF) integrating Huber's M-estimation with CKF.
• To enhance state estimation accuracy in nonlinear systems with missing data.

Main Methods:

• Integration of Huber's M-estimation theory into the conventional cubature Kalman filter (CKF).
• Utilizing covariance matrix predictions for state estimation and control amidst dynamic errors.
• Comparative evaluation against Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), and CKF.

Main Results:

• The RCKF demonstrated superior performance in handling anomaly errors compared to EKF, EnKF, and CKF.
• Significantly lower Average Root Mean Square Error (ARMSE) and Average Normalized Cross-correlation Integral (ANCI) were achieved with RCKF.
• Consistent performance improvements observed in both Univariate Non-stationary Growth Model (UNGM) and bearing-only tracking (BOT) examples.

Conclusions:

• The proposed RCKF effectively mitigates performance degradation caused by missing measurements and noise deviations.
• RCKF offers a robust solution for state estimation in nonlinear systems facing data loss and dynamic disruptions.
• RCKF shows significant advantages over existing Kalman filter variants in challenging tracking scenarios.