ON A CLASS OF STOCHASTIC PURSUIT - EVASION GAMES
HARVARD UNIV CAMBRIDGE MA CAMBRIDGE United States
Pagination or Media Count:
The report continues the investigation of a class of stochastic differential games introduced in an earlier report by Willman AD-679 232. Iterative algorithms for computing numerical solutions to games of this type are discussed. It is shown by numerical example that there exist multistage games analogous to this type of differential game for which minimax solutions do indeed exist. Several candidates for quasi-optimal strategies are presented which are simple to compute and easy to implement compared to the minimax strategies. A criterion is developed for evaluating the performance of non-optimal strategies. The performances of these quasi-optimal strategies are evaluated in terms of this criterion for a numerical example an interception problem involving two second-order systems.
- Operations Research