A Generalized Labeled Multi-Bernoulli Filter for Maneuvering Targets

TL;DR

This paper introduces a Generalized Labelled Multi-Bernoulli filter designed for tracking maneuvering targets modeled as Jump Markov Systems, validated through linear and nonlinear tracking examples.

Contribution

It develops a novel GLMB filter specifically for maneuvering targets with Markovian motion models, enhancing tracking accuracy in complex scenarios.

Findings

Effective in linear and nonlinear maneuvering target tracking

Improves tracking performance over existing methods

Validated with multiple simulation examples

Abstract

A multiple maneuvering target system can be viewed as a Jump Markov System (JMS) in the sense that the target movement can be modeled using different motion models where the transition between the motion models by a particular target follows a Markov chain probability rule. This paper describes a Generalized Labelled Multi-Bernoulli (GLMB) filter for tracking maneuvering targets whose movement can be modeled via such a JMS. The proposed filter is validated with two linear and nonlinear maneuvering target tracking examples.