A robust pre-filtering approach to EKF underwater target tracking

Abstract

A robust approach to solving the passive underwater target tracking problem based on the extended Kalman filtering (ERF) is proposed in this paper. The conventional method based on the assumption of Gaussian noise statistics is not robust in many instances and the resulting filter is likely to diverge even for the slightest deviation from the Gaussian assumption. The proposed approach involves pre-processing of data using a robust M-estimate pre-filter. Monte Carlo simulation results for test cases involving heavy-tailed contaminated observation noise demonstrate the robustness of the proposed estimation procedure.<>