Robust Particle Filter by Dynamic Averaging of Multiple Noise Models

TL;DR

This paper introduces a robust particle filter that dynamically combines Gaussian and Student's t noise models using Bayesian averaging, improving state estimation accuracy in the presence of measurement outliers.

Contribution

It proposes a novel particle filter method that adaptively weights multiple noise models, enhancing robustness against outliers compared to traditional approaches.

Findings

Demonstrates improved robustness in simulations with outliers

Automatically adjusts noise model influence during filtering

Outperforms standard particle filters in noisy scenarios

Abstract

State filtering is a key problem in many signal processing applications. From a series of noisy measurement, one would like to estimate the state of some dynamic system. Existing techniques usually adopt a Gaussian noise assumption which may result in a major degradation in performance when the measurements are with the presence of outliers. A robust algorithm immune to the presence of outliers is desirable. To this end, a robust particle filter (PF) algorithm is proposed, in which the heavier tailed Student's t distributions are employed together with the Gaussian distribution to model the measurement noise. The effect of each model is automatically and dynamically adjusted via a Bayesian model averaging mechanism. The validity of the proposed algorithm is evaluated by illustrative simulations.