Marginal multi-Bernoulli filters: RFS derivation of MHT, JIPDA and association-based MeMBer
This work addresses multi-target tracking for applications like surveillance or robotics, but it is incremental as it builds on existing RFS methods to derive known algorithms.
The paper tackles the problem of multi-target tracking by deriving a full Bayes random finite set filter that reveals implicit data association, similar to MHT, and develops two algorithms approximating association distributions. The results show improved performance in challenging environments, with algorithms nearly identical to JIPDA and related to the MeMBer filter.
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.