Comparison of interactive multiple model particle filter and interactive multiple model unscented particle filter for tracking multiple manoeuvring targets in sensors array

Tracking multiple targets in cluttered environment has been acknowledged as a challenging task involving handling of measurement track-to-track uncertainty association in conjunction with nonlinearity and imprecision pervading the target dynamic models. In this paper an approach based on the use of an interacting multiple model particle filter (IMMPF) has been put forward, where the particle filter (PF) allows the system to handle non-linearity of the target cinematic models while the interacting multiple model (IMM) deals with the model switch when a target changes its manoeuvre. On the other hand, Cheap Joint Probabilistic Data Association (CJPDA) was used to tackle the data association problem. Two fusion architectures using the federated and the centralized form of Kalman filter were investigated. Performances and feasibility of the proposal are demonstrated through a set of Monte Carlo simulations involving three crossing targets. Also, a comparison analysis with an alternative approach using the IMM filter in conjunction with the Unscented Particle Filter (IMMUPF) is carried out. The results demonstrate the feasibility of the proposal and satisfactory tracking of the targets.