Fuzzy assisted Kalman filtering for target tracking

This paper is concerned with estimating the states of a constant velocity crossing target. A new approach to estimation of a nonlinear system is outlined, which uses a neurofuzzy system model to produce a linear state equation. This has the immediate advantage of enabling the linear Kalman filter to be used on systems which may be highly nonlinear. Results from simulation studies comparing the fuzzy assisted Kalman filter with conventional extended Kalman filtering, in a target tracking application demonstrate the efficacy of this new technique.