Marginal multi-Bernoulli filters: RFS derivation of MHT, JIPDA and association-based MeMBer

TL;DR

This paper derives a full Bayesian RFS filter revealing implicit data association, leading to algorithms akin to JIPDA and MeMBer that enhance multi-target tracking performance in difficult scenarios.

Contribution

It introduces a new RFS-based derivation that unifies data association methods like JIPDA and MeMBer within a Bayesian framework.

Findings

Algorithms improve tracking accuracy in challenging environments

Derived filters closely resemble existing methods JIPDA and MeMBer

Provides a unified RFS-based perspective on data association techniques

Abstract

Recent developments in random finite sets (RFSs) have yielded a variety of tracking methods that avoid data association. This paper derives a form of the full Bayes RFS filter and observes that data association is implicitly present, in a data structure similar to MHT. Subsequently, algorithms are obtained by approximating the distribution of associations. Two algorithms result: one nearly identical to JIPDA, and another related to the MeMBer filter. Both improve performance in challenging environments.