Abstract:This paper deals with state estimation of nonlinear stochastic dynamic models. In particular, the stochastic integration rule, which provides asymptotically unbiased estimates of the moments of nonlinearly transformed Gaussian random variables, is reviewed together with the recently introduced stochastic integration filter (SIF). Using SIF, the respective multi-step prediction and smoothing algorithms are developed in full and efficient square-root form. The stochastic-integration-rule-based algorithms are implemented in Python (within the Stone Soup framework) and in MATLAB and are numerically evaluated and compared with the well-known unscented and extended Kalman filters using the Stone Soup defined tracking scenario.

What problem does this paper attempt to address?

### Problems the paper attempts to solve This paper aims to solve the state estimation problem of nonlinear stochastic dynamic models. Specifically, the author focuses on how to provide asymptotically unbiased moment estimates when dealing with Gaussian random variables with nonlinear transformations. To this end, the paper reviews the method based on the Stochastic Integration Rule (SIR) and introduces the recently proposed Stochastic Integration Filter (SIF). The main objectives of the paper are: 1. **Design multi - step prediction and smoothing algorithms**: Complete the design of multi - step prediction and smoothing algorithms in the SIF family. 2. **Develop efficient and robust estimators**: Design efficient estimators in the square - root form to improve numerical stability and computational efficiency. 3. **Implement and evaluate new algorithms**: Implement the latest SIF and its related prediction and smoothing algorithms in Python (using the Stone Soup framework) and MATLAB, and conduct numerical evaluation and comparison through the tracking scenarios defined by Stone Soup. ### Key technical background - **State estimation problem**: State estimation of discrete - time stochastic dynamic systems extracts system state information from noisy or incomplete measurements and has important applications in fields such as navigation, speech and image processing, fault detection, optimal control, and tracking. - **Bayesian method**: The general state estimation solution is given by Functional Recursive Relations (FRRs) for calculating the conditional probability density function (PDF), but these relations usually have analytical solutions only for linear models. - **Gaussian estimator (GE)**: For nonlinear models, common approximation methods include the Extended Kalman Filter (EKF), the Unscented Kalman Filter (UKF), etc. - **Stochastic integration filter (SIF)**: Compared with other Gaussian filters, SIF provides asymptotically exact integral evaluation, thus improving the estimation performance. ### Main contributions 1. **Multi - step prediction and smoothing algorithms**: Proposed multi - step prediction and smoothing algorithms based on SIF, filling the gap in this area. 2. **Efficient square - root form**: Developed an efficient square - root form of SIF, improving numerical stability and computational efficiency. 3. **Public implementation**: Implemented SIF and its related algorithms in the Stone Soup framework of Python and MATLAB, and conducted numerical evaluation, proving its superior performance. ### Numerical experiment results By conducting experiments on a specific tracking scenario, the paper shows the superior performance of SIF compared with EKF and UKF. Specifically: - **Root Mean Square Error (RMSE)**: The RMSE performance of SIF on all state variables is very close to that of EKF and UKF. - **Average Normalized Estimation Error Squared (ANEES)**: The ANEES of SIF is significantly better than that of EKF and UKF, with an improvement of about 40% respectively. In summary, this paper significantly improves the state estimation performance of nonlinear stochastic dynamic models by proposing new multi - step prediction and smoothing algorithms and an efficient square - root form of SIF, and verifies its superiority through experiments.

Stochastic Integration Based Estimator: Robust Design and Stone Soup Implementation

Input and state estimation for linear systems: A least squares estimation approach

State Estimation For Nonlinear Systems By Using Dynamic Optimization

Distributed Robust Bayesian Filtering for State Estimation.

State Estimation for Continuous-Discrete-Time Nonlinear Stochastic Systems

Robust State Estimation for Uncertain Linear Systems with Random Parametric Uncertainties

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

Robust State Estimation for Uncertain Linear Systems with Deterministic Input Signals

Optimization based data enrichment using stochastic dynamical system models

Stochastic Nonlinear Model Predictive Control with State Estimation by Incorporation of the Unscented Kalman Filter

Effective filtering approach for joint parameter-state estimation in SDEs via Rao-Blackwellization and modularization

Robust State Estimation for Uncertain Discrete-Time Stochastic Systems with Missing Measurements

A Nonlinear Stochastic Filter for Continuous-Time State Estimation

A Concentration-Based Approach for Optimizing the Estimation Performance in Stochastic Sensor Selection

l1 norm‐based recursive estimation for non‐linear systems with non‐Gaussian noises

Design of Efficient Point-Mass Filter with Application in Terrain Aided Navigation

Sigma-point Kalman Filter with Nonlinear Unknown Input Estimation via Optimization and Data-driven Approach for Dynamic Systems

Wasserstein Distributionally Robust Control and State Estimation for Partially Observable Linear Systems

UD-based pairwise and MIMO Kalman-like filtering for estimation of econometric model structures

Stochastic parameterization identification using ensemble Kalman filtering combined with expectation-maximization and Newton-Raphson maximum likelihood methods