Robust Range-Only Tracking via Angle-Parameterized Maximum Correntropy Square-Root Cubature Kalman Filtering

Abstract

Range-only tracking is a critical nonlinear filtering problem in surveillance applications where angle information is unavailable or unreliable. Standard Kalman-based filters suffer from significant performance degradation under two practical challenges: Large initial angular uncertainty and heavy-tailed measurement noise. To address both issues simultaneously, a novel angle-parameterized maximum correntropy square-root cubature Kalman filter (AP-MCSCKF) is proposed in this paper. The proposed method adopts an angle-parameterized initialization strategy. The initial angular uncertainty region is partitioned into multiple subintervals, and a bank of parallel MCSCKF sub-filters is initialized with different initial angles. To suppress the influence of impulsive outliers, a correntropy-based pseudo-likelihood is introduced to update the sub-filter weights according to the whitened measurement innovations. This design inherits numerically stable advantages of the square-root cubature Kalman filter while providing enhanced robustness to both initialization errors and measurement outliers. Simulation results show that, under Gaussian measurement noise, the proposed AP-MCSCKF achieves performance comparable to that of the SCKF while retaining the benefit of angle-parameterized initialization. Under outlier-contaminated non-Gaussian measurement noise, the proposed method achieves improved robustness and steady-state tracking accuracy while maintaining good filter consistency.