Incomplete Information Pursuit-Evasion Games with Uncertain Relative Dynamics

Abstract

Pursuit-evasion games are multiple player differential games that are commonly posed as optimal control problems. In a complete information game, each player has access to the true gains and control input associated with the other players. A more realistic scenario involves an incomplete information game where players have no knowledge of their opponent strategies. Behavior learning methods can be used to combat incomplete information games and allow a player to gain a tactical advantage. The addition of uncertain relative dynamics further complicates these pursuit-evasion scenarios because the true dynamic model is used in the optimal control solution and to estimate opponent behavior. A technique for incomplete information games with uncertain relative dynamics is presented. This method is applied to a numerical example and the results are compared to a game with certain relative dynamics.