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.

Robust Gaussian-sum ensemble Kalman filter and its application in bearings-only tracking

An Improved Gaussian Mixture Particle Filter Based Targets Tracking Algorithm for Bearing-Only Tracking System

Robust SLAM using square-root cubature Kalman filter and Huber’ s GM-estimator

Gaussian Particle Filter Based Pose and Motion Estimation

Robust Cubature Kalman Filter for Moving-Target Tracking with Missing Measurements

An Adaptive Kalman Filtering Algorithm for Autonomous Orbit Determination Based-on BP Neural Network

A Monte Carlo-Based Iterative Extended Kalman Filter for Bearings-Only Tracking of Sea Targets

Huber-Based Adaptive Unscented Kalman Filter with Non-Gaussian Measurement Noise

Bearings-only Tracking Based on Distributed Multisensor Pseudolinear Kalman Filter

Robust Kalman Filters Based on the Sub-Gaussian $α$-stable Distribution

Adaptive Two-Step Filter with Applications to Bearings-Only Measurement Problem

Design of robust Gaussian approximate filter and smoother with latency probability identification

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

A Novel Adaptive Robust Cubature Kalman Filter for Maneuvering Target Tracking with Model Uncertainty and Abnormal Measurement Noises

A comparison of nonlinear extensions to the ensemble Kalman filter: Gaussian Anamorphosis and Two-Step Ensemble Filters

Variational Bayesian-Based Robust Cubature Kalman Filter With Application on SINS/GPS Integrated Navigation System

A novel dynamic outlier-robust Kalman filter with Moving Horizon Estimation

Robust Cubature Kalman Filter for SINS/GPS Integrated Navigation Systems With Unknown Noise Statistics

A Pseudolinear Maximum Correntropy Kalman Filter Framework for Bearings-only Target Tracking

Credible Gaussian sum cubature Kalman filter based on non-Gaussian characteristic analysis

Nonlinear Bayesian Filtering with Natural Gradient Gaussian Approximation