Multi mode projectile tracking with Marginalized Particle Filter

Abstract

In this study, dynamic models for thrusting and ballistic flight modes of multi mode projectile are obtained and Marginalization method is applied by separation of the linear and nonlinear parts of state space model. In Marginalized Particle Filter (MPF), dimension of the nonlinear system is reduced so that the model can be utilized to obtain better estimates of the state using the same number of particles as that of standard particle filter. The Extended Kalman Filter (EKF), the Particle Filter (PF) and the Marginalized Particle Filter (MPF) are compared by their RMS errors in position and velocity estimations obtained by Monte Carlo simulations. In general, EKF has the best performance on position estimation and MPF has the best performance on velocity estimation.