A consistency-based gaussian mixture filtering approach for the contact lens problem

In this paper a novel consistency-based Gaussian mixture nonlinear filter (CbGMF) is proposed where the distribution of the target state is represented by a dynamic set (mixture) of Gaussian distributions (“subtracks”). The subtracks are generated using a consistency-based filtering rule for the EKF and a novel approach for consistent track splitting. Simulation results show that the CbGMF has performance superior to previous algorithms for a tracking problem with a contact lens shaped uncertainty in the state estimation error as well as in keeping the range estimation error small in the early stages of the filtering.