An Improvement of Ensemble Kalman Filter for OOSM Tracking

Abstract In multisensor target tracking system, the measurements of the same target are always delayed, come at different rates, and arrive out of sequence. Such measurements are called “out-of sequence” measurements (OOSMs). Examples of such systems are a mobile robot or an unmanned aerial vehicle (UAV) which is observed by both inertial sensors and visual sensors, and delay caused by transmitting or processing time. Solutions via Extended Kalman Filter (EKF), Particle Filter (PF), and Ensemble Kalman Filter (EnKF) have been proposed so far. EnKF yields various advantages, e.g., it can be applied to strong nonlinear system, requires much less particles than PF, and does not require any Jacobian matrix or backward state-transition function. In this paper, we propose an algorithm to improve accuracy of using EnKF with OOSMs. The algorithm concerns estimating an ensemble of the process noises. We validate the proposed algorithm by simulations of the aircraft tracking system. The results show another competitive solution for OOSM filtering.