Impact Time and Point Predicted Using a Neural Extended Kalman Filter

Predictive target tracking is becoming increasingly important for a wide variety of applications, including for the prediction of when and where a ballistic projectile may hit the ground. Not only is such predictive tracking useful for providing a determination of whether the target will strike a valuable asset, it can also provide information about the order in which targets may be struck, and provide for the most effective use of resources in responding to threats. One such problem is the tracking of a ballistic trajectory using information from a ground-based sensor to direct high-cost precision strike munitions. A neural Kalman filter approach to target tracking is presented as a technique to improve the motion model of the target while it is being tracked in flight. A linearized version of that model is then used to provide an improved estimate of the predicted location of the target. Results are presented from use of a neural extended Kalman filter for predictive target tracking of a ballistic trajectory. The tracker is able to fuse information over the course of the trajectory and, as a result, estimate the time and location that the ballistic target will strike. To respond to manoeuvres by the target, the motion model becomes a composite of the a priori motion model of a ballistic trajectory and a neural network. Prediction to ground impact is calculated and updated throughout the trajectory after each sensor measurement.