Information-Based Particle Flow With Convergence Control

Abstract

A new formulation of the Gaussian particle flow filter is presented using an information theoretic approach. The developed information-based form advances the Gaussian particle flow framework in two ways: it imparts physical meaning to the flow dynamics and provides the ability to easily include modifications for a non-Bayesian update. An orbit determination simulation with high initial uncertainty is used to demonstrate the consistent, robust performance of the information flow filter in situations where the extended Kalman filter fails.