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.

Building Nonlinear Filter Toolbox with Scilab

A New Framework for Nonlinear Kalman Filters

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

Nonlinear Bayesian Filtering with Natural Gradient Gaussian Approximation

Solving Nonlinear Filtering Problems in Real Time by Legendre Galerkin Spectral Method.

Continuous-discrete unscented Kalman filtering framework by MATLAB ODE solvers and square-root methods

Kalman/particle Filter's Design and Simulation in LSINS/TAN

On derivative-free extended Kalman filtering and its Matlab-oriented square-root implementations for state estimation in continuous-discrete nonlinear stochastic systems

Performance Comparison of Various Suboptimal Nonlinear Kalman Filter Methods for Electro-Optical Tracking System

A tutorial on particle filters for online nonlinear/non-Gaussian Bayesian tracking

The Geometric Unscented Kalman Filter

Algorithm of Gaussian Sum Filter Based on SGQF for Nonlinear Non-Gaussian Models

Gaussian Sum Filtering Methods for Nonlinear Non-Gaussian Models

Application of a New Nonlinear Filtering Method in Satellite Attitude Estimation

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

Gaussian non-linear filter algorithm based on integrated navigation system

Algorithm of Gaussian Sum Based Cubature Kalman Filter for Non-Gaussian Systems

A New Linear Regression Kalman Filter with Symmetric Samples.

Hermite Spectral Method to 1-D Forward Kolmogorov Equation and Its Application to Nonlinear Filtering Problems

Kalman Filter, Unscented Filter and Particle Flow Filter on Non-linear Models

Parameter-free and fast nonlinear piecewise filtering. Application to experimental physics