Sensitivity reduction of a class of zero-sum two-player linear deterministic differential games

The problem of reducing the terminal state sensitivity to small variations in plant parameters for a class of zero-sum two-player linear differential games with a fixed target was examined by implementing the optimal strategies in a combined open-loop and closed-loop form. A direct search computational procedure was adapted for selection of two feedback matrices used in the strategies, if the elements in the matrices are restricted to have constant values. A specific example of a linear fixed-target pursuit-evasion game was illustrated. It was noted that not only the terminal state sensitivities can be effectively reduced to zero but the state sensitivities at all times until termination of the game are also greatly reduced.