Research on Statistical Similarity Measure-based Outlier-Robust Fusion Method for Underwater Navigation

It is crucial for an underwater vehicle to gather precise location data. However, the measurement outliers caused by complex underwater disturbances make it difficult to obtain precision state estimation. M-estimate is a classical method for processing measurement outliers, however, it lacks the ability to maintain both steady-state estimation accuracy and rapid convergence at the same time. To address the problem, this paper proposes a novel fusion method, and a new cost function is created by integrating the measurement of statistical similarity into the framework of M-estimate. Simulations are performed to assess how well the suggested algorithm works. The simulation results indicate that the proposed algorithm delivers enhanced state estimation precision and robustness when compared with other methods.