Improved Rao-Blackwellized particle filtering algorithms for multi-target tracking in clutter

We consider the problem of multiple target tracking in the presence of clutter (false alarm) measurements. To improve the performance of the Rao-Blackwellized particle filter (RBPF) data association algorithms, some simple but effective strategies are implemented. We first present a sequential likelihood method, i.e., all measurements are used to update the particles more than one time in each time step. It is observed that the tracking performance of the algorithm is not severely loss with fewer particles. We then present a simple gating technique to reduce the validated measurements to a feasible level. It is worth mentioning that the association probabilities are not calculated by grouping targets into clusters as the joint probabilistic data association (JPDA), but only reserve the validated measurements in the joint validation region (gate) and ignore the measurements outside. Simulations are also presented to compare the performance of the proposed algorithms.