Coordinate ascent neural Kalman-MLE for state estimation

TL;DR

This paper introduces a coordinate ascent algorithm that learns neural network-based dynamic and measurement models for state estimation, optimizing model parameters and noise covariances via maximum likelihood for improved filtering accuracy.

Contribution

It proposes a novel supervised learning approach combining neural networks with Kalman filtering for dynamic state estimation.

Findings

Effective learning of neural dynamic and measurement models.

Improved state estimation accuracy with learned models.

Joint optimization of model parameters and noise covariances.

Abstract

This paper presents a coordinate ascent algorithm to learn dynamic and measurement models in dynamic state estimation using maximum likelihood estimation in a supervised manner. In particular, the dynamic and measurement models are assumed to be Gaussian and the algorithm learns the neural network parameters that model the dynamic and measurement functions, and also the noise covariance matrices. The trained dynamic and measurement models are then used with a non-linear Kalman filter algorithm to estimate the state during the testing phase.