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.

Nonlinear/Non-Gaussian Bayesian Sequential Filtering

Nonlinear Bayesian Filtering with Natural Gradient Gaussian Approximation

Gaussian Sum Filtering Methods for Nonlinear Non-Gaussian Models

The discriminative Kalman filter for nonlinear and non-Gaussian sequential Bayesian filtering

On sequential Monte Carlo sampling methods for Bayesian filtering

Multisensor Sequential Particle Filter

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

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

Ensemble Kalman filter based sequential Monte Carlo sampler for sequential Bayesian inference

Distributed Robust Bayesian Filtering for State Estimation.

Closed-form Filtering for Non-linear Systems

Estimating States for Nonlinear Systems Using the Particle Filter

A Nonlinear Stochastic Filter for Continuous-Time State Estimation

Perron-Frobenius operator filter for stochastic dynamical systems

[Re] The Discriminative Kalman Filter for Bayesian Filtering with Nonlinear and Non-Gaussian Observation Models

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

Recursive filtering with non-Gaussian noises

State Estimation for Nonlinear Discrete-Time Fractional Systems: A Bayesian Perspective

A Progressive Bayesian Filtering Framework for Nonlinear Systems With Heavy-Tailed Noises

Perron–Frobenius Operator Filter for Stochastic Dynamical Systems

Variational Bayesian Adaptation of Noise Covariances in Non-Linear Kalman Filtering