An improved PMHT using an idea from coding

Tracking is inherently a combinatorial optimization problem under the (admittedly realistic) constraint that each target generates at most one measurement per scan per sensor. Practical algorithms to solve the combinatorial problem are usually intelligent suboptimal procedures. Optimal procedures can be derived if the constraint above is relaxed. The PMHT (probabilistic multi-hypothesis tracker) uses “soft” posterior-probability associations between measurements and targets. Its implementation is a straightforward iterative application of the Kalman smoother operating on “synthetic” (i.e., modified) measurements, and of recalculation of these synthetic measurements based on the current track estimate. As applied to data fusion the PMHT is a very natural procedure, in that complexity is generally linear in the number of sensors. In this presentation, we first discuss the basic PMHT and some of the older PMHT variants which have been used to enhance convergence. We then treat a new turbo-PMHT, which is informed by the recent success of turbo coding in communication contexts. This new PMHT has performance substantially improved versus any of the previous versions, and generally as good as the probabilistic data association filter