Control Strategies for Players with Discrete and Uncertain Observations

In this paper, robust multiplayer games are generalized from a deterministic scenario with known multiplayer dynamics to a stochastic scenario where each player has uncertain dynamics and estimates other players’ dynamics through uncertain observations. A design approach to compute control strategy for each player given those observations, is provided. Information from other players is assumed to be available to each player only at discrete time instances and is assumed to be corrupted by noise. Having available only this limited and somewhat corrupted information due to its dependence on noise, each player has to make their own decisions based on estimated states of other players which results in scenarios that may range from cooperative to noncooperative. Decisions are integrated into designs of most appropriate actions, that is, control strategies of each player. The design is illustrated on a multiplayer pursuit-evasion simulation example.