Predicting Pop up Threats from an Adaptive Markov Model

Abstract

We consider military scenarios in which an adversary has ground assets that are not observable most of the time, and on occasion, they pop up and become observable for short durations of time. It is desired to deploy Unmanned Air Vehicles and persistently deny the adversary time windows for launching weapons. We model the pop up phenomenon as an Adaptive Markov Chain, and use the model to predict future pop up threat locations. The UAVs are controlled to move towards the predicted threats to better perform Persistent Area Denial Mission. The objective is to reduce the time between the pop up and the time to reach the pop up. Preliminary simulation experiments are presented.