Randomized unscented Kalman filter in target tracking

摘要

The paper deals with state estimation of nonlinear non-Gaussian stochastic dynamic systems with a special focus on target tracking. The randomized unscented Kalman filter is introduced. The filter utilizes the randomized unscented transform, which is based on a degree 3 stochastic integration rule. The paper discusses several aspects of the randomized unscented transform and its relation to the unscented transform, which can be seen as a special deterministic version. The aspects are demonstrated on polar to Cartesian coordinate transformation. The randomized unscented Kalman filter is illustrated in a numerical example concerning a reentry problem.