Computing the Optimal Mixed Strategy for Various Ambush Games

摘要

This paper studies a variety of ambush games in which one player must navigate between an origin and a destination, and the other player seeks to intercept and ambush him. These games include single-stage games, in which all decisions are made at the outset of the game, and multi-stage games, in which the second player may choose his ambush locations based on real-time updates of the first player's position. For both types of ambush games, methods are presented for efficiently computing the optimal mixed strategies for the first player to navigate between the origin and destination, so as to minimize the probability of being ambushed.