Multisensor fusion estimation of nonlinear systems with intermittent observations and heavy-tailed noises

Inspired by the robust student t-distribution based nonlinear filter (RSTNF), a student t-distribution and unscented transform (UT) based filter for state estimation of heavy-tailed nonlinear dynamic systems, a modified RSTNF for intermittent observations is derived. The fusion estimation for nonlinear multisensor systems with intermittent observations and heavy-tailed measurement and process noises is studied. In this work, the centralized fusion, the sequential fusion, and the naïve distributed fusion algorithms are presented, respectively. Theoretical analysis shows that the presented algorithms are effective, which are the efficient extension of the classical unscented Kalman filter (UKF) or the cubature Kalman filter (CKF) based algorithms with Gaussian noises. Simulation results show that the presented algorithms are effective and feasible.