Approximate Bayesian Smoothing with Unknown Process and Measurement Noise Covariances

TL;DR

This paper introduces an adaptive Bayesian smoothing method for linear systems with unknown noise covariances, using variational Bayes for efficient approximate inference, demonstrated on a target tracking task.

Contribution

It proposes a novel variational Bayes-based adaptive smoother that handles unknown noise covariances in high-dimensional linear models.

Findings

Efficient and easy-to-implement algorithm

Applicable to high-dimensional systems

Effective in target tracking scenarios

Abstract

We present an adaptive smoother for linear state-space models with unknown process and measurement noise covariances. The proposed method utilizes the variational Bayes technique to perform approximate inference. The resulting smoother is computationally efficient, easy to implement, and can be applied to high dimensional linear systems. The performance of the algorithm is illustrated on a target tracking example.