A multisensor data fusion-based target tracking system

Abstract

In this paper a multisensor data fusion-based target tracking system is presented. The system includes neuro-fuzzy multisensor data fusion in order to overcome the limitation of the use of a single sensor. It has the capability of minimising the noise that contaminates the sensor measurements and excludes the faulty (invalid) measurements from use in the estimation process. Despite being a simple algorithm, it can deal with the data fusion problem using noisy nonlinear sensors as well as linear sensors. A neuro-fuzzy kinematics process model is also employed in this target tracking system to cope with the lack of a priori knowledge of the target dynamics. Although no a priori statistical knowledge of the target dynamics and the sensors are involved in the estimation process, the performance of the proposed system is comparable with the extended Kalman filter-based target tracking system which uses the exactly known process model of the target.