Application of Recurrent Neural-Network based Kalman Filter for Uncertain Target Models

For various target tracking applications, it is well known that the Kalman filter optimal estimator(in minimum mean-square sense) to predict and estimate state(position and/or velocity) of linear dynamical systems driven by Gaussian stochastic noise. In case nonlinear systems, Extended filter(EKF) Unscented filter(UKF) are widely used, which can be viewed as approximations the(linear) in sense conditional expectation. However, implement EKF UKF, exact model information statistical noise still required. this paper, we propose recurrent neural-network based filter, where its gain obtained via proposed GRU-LSTM framework does not need precise covariance information. By state estimation performance enhanced terms error, verified through problems with incomplete