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.

Adaptive Gaussian Mixture Filter Based on Statistical Linearization

Gaussian mixture density modeling and decomposition with weighted likelihood

A Bayesian Filtering Algorithm for Gaussian Mixture Models

Gaussian Mixture Multiple-Model Multi-Bernoulli Filters For Nonlinear Models Via Unscented Transforms

Variational Adaptive Kalman Filter With Gaussian-Inverse-Wishart Mixture Distribution

An Adaptive Covariance Parameterization Technique for the Ensemble Gaussian Mixture Filter

A Gaussian-Multivariate Laplacian Mixture Distribution Based Robust Cubature Kalman Filter

Approximate MMSE Estimator for Linear Dynamic Systems With Gaussian Mixture Noise

Gaussian implementation of the multi-Bernoulli mixture filter

A Novel Robust Gaussian–Student's t Mixture Distribution Based Kalman Filter

A Gaussian approximation filter for nonlinear systems with compound noises

Adaptive Gaussian Mixture Model-Based Relevant Sample Selection for JITL Soft Sensor Development

Adaptive Gaussian mixture filter for Markovian jump nonlinear systems with colored measurement noises

A Hessian for Gaussian Mixture Likelihoods in Nonlinear Least Squares

Nonlinear Bayesian Filtering with Natural Gradient Gaussian Approximation

Online Adaptive Statistical Compressed Sensing of Gaussian Mixture Models

Gating technique for the Gaussian mixture multi-Bernoulli filter

Particle Gaussian Mixture (PGM) Filters

Variational Gaussian filtering via Wasserstein gradient flows

A hybrid particle-ensemble Kalman filter for problems with medium nonlinearity

Robust Cubature Kalman Filter With Gaussian-Multivariate Laplacian Mixture Distribution and Partial Variational Bayesian Method