Interacting Methods for Manoeuvre Handling in the GM-PHD Filter

Probability hypothesis density (PHD) filter implementations with jump Markov linear system (JMLS) manoeuvre handling are well known. A new derivation, following the method of the popular interacting multiple model (IMM) filter from single target tracking, is presented and is shown to be equivalent to existing implementations. The resulting filter is shown to reduce to the optimal Bayes solution for jointly estimating target state and motion mode in the single target, linear, and Gaussian case.