framework of joint optimization algorithm based on EM (Expectation Maximization) is proposed for tracking a boost-phase ballistic target with unknown ballistic parameters.Firstly
the state and unknown parameters are estimated based on smoother in the E step.Then the mean and covariance of initial states
and the noise covariances are calculated in the M step.At last
URTS (Unscented Rauch-Tung-Striebel) based on EM is derived and the analytical forms of unknown statistics parameters are given
which makes the non-convex numerical optimization unnecessary.The result shows that the proposed algorithm is more accurate than iterative UKF (Unscented Kalman Filter) with the same order of magnitude of calculation.