APPLICATION OF NEURAL NETWORKS IN TARGET TRACKING DATA FUSION

Kalman filtering is a fundamental building block of most multiple-target tracking (MTT) algorithms. The other building block usually involves some type of data association schemes. Here it is proposed to incorporate a neural network into the normal Kalman filter configuration such that the neural network provides the adaptive capability the filter needs. As such, the estimation error of the Kalman filter would be reduced, hence improving the MTT solution. Simulation results have shown that this claim is valid.