Simultaneous Search and Monitoring of Evasive Targets by an Unmanned Aerial Vehicle

Abstract

This paper studies a Simultaneous Search and Monitoring (SSM) problem between an aerial pursuer and multiple evasive ground targets. The adversarial target behaviours and the partial observation by the pursuer make this problem a Partially Observable Stochastic Game (POSG). Because of the intractability of obtaining an exact solution, we applied some assumptions on the information available to the targets. These assumptions simplify the game playing into a dynamic Guaranteed Search problem. A policy planning algorithm was designed for the pursuer, which achieved efficient Simultaneous Search and Monitoring and reduced the total uncertainty level. The comparative study validated the advantages of our policy planning over conventional fixed-pattern Guaranteed Search method.