Abstract:Practical Bayes filters often assume the state distribution of each time step to be Gaussian for computational tractability, resulting in the so-called Gaussian filters. When facing nonlinear systems, Gaussian filters such as extended Kalman filter (EKF) or unscented Kalman filter (UKF) typically rely on certain linearization techniques, which can introduce large estimation errors. To address this issue, this paper reconstructs the prediction and update steps of Gaussian filtering as solutions to two distinct optimization problems, whose optimal conditions are found to have analytical forms from Stein's lemma. It is observed that the stationary point for the prediction step requires calculating the first two moments of the prior distribution, which is equivalent to that step in existing moment-matching filters. In the update step, instead of linearizing the model to approximate the stationary points, we propose an iterative approach to directly minimize the update step's objective to avoid linearization errors. For the purpose of performing the steepest descent on the Gaussian manifold, we derive its natural gradient that leverages Fisher information matrix to adjust the gradient direction, accounting for the curvature of the parameter space. Combining this update step with moment matching in the prediction step, we introduce a new iterative filter for nonlinear systems called \textit{N}atural Gr\textit{a}dient Gaussia\textit{n} Appr\textit{o}ximation filter, or NANO filter for short. We prove that NANO filter locally converges to the optimal Gaussian approximation at each time step. Furthermore, the estimation error is proven exponentially bounded for nearly linear measurement equation and low noise levels through constructing a supermartingale-like property across consecutive time steps.

A new method for the nonlinear transformation of means and covariances in filters and estimators

Weighted Minimum Mean Square Kalman Filter

A New Framework for Nonlinear Kalman Filters

A New Linear Regression Kalman Filter with Symmetric Samples.

A New computation reduction based nonlinear Kalman filter

Convergence Analysis of Nonlinear Kalman Filters with Novel Innovation-Based Method

A New Nonlinear State Estimator Using the Fusion of Multiple Extended Kalman Filters

Kalman Filter and its Modern Extensions for the Continuous-time Nonlinear Filtering Problem

A Gaussian Nonlinear Iterated Update Filter

Nonlinear Kalman Filtering with Reparametrization Gradients

Nonlinear Kalman Filtering with Divergence Minimization

Adaptive Error Covariances Estimation Methods for Ensemble Kalman Filters

Nonlinear Bayesian Filtering with Natural Gradient Gaussian Approximation

A Method of Approximate Second-Order Extended Kalman Filter

Kernel Based Filter for Nonlinear Estimation.

Distributed fusion cubature Kalman filters for nonlinear systems

Survey of Nonlinear Filters in the Framework of Recursive Bayesian Estimation

A New Method Based on the Polytopic Linear Differential Inclusion for the Nonlinear Filter

An Improved Iterative Maximum Correntropy Cubature Kalman Filter

A Novel Adaptive Unscented Kalman Filter for Nonlinear Estimation

Iterated Kalman Methodology For Inverse Problems