Set-membership PHD filter for multiple maneuvering targets

Abstract:
This paper proposes a multiple model probability hypothesis density (MM-PHD) filter for multiple maneuvering targets based on a linear dynamic system in which the initial state, process and measurement noises are known by representing compact sets (e.g., polytopes).The setting of set-membership MM-PHD (SM-MM-PHD) is the same as the standard MMPHD. We derive the equations of the prediction and update process of SM-MM-PHD in terms of set of probability measures. Compared to the standard MM-PHD, this proposed SM-MM-PHD can guarantee the targets' states based on compact sets. The performance of the proposed SM-MM-PHD algorithm is evaluated by means of simulation experiments.