An iterative learning process based on Bayesian principle in pursuit-evasion games

In a pursuit-evasion game, a pursuer tries to capture evader while the evader is trying to avoid capture. During the pursuit and evasion, pursuer longs for minimizing the distance from evader at any time while evader wants to the maximal distance. Although the relevant information of each side is unknown for each other, the initial information about pursuer and evader's locations and transition directions can be presented according to the prior probability. Then a Bayesian iterative process can be used to modify the probability of opponent's actions and to maximize the probability. It can make the pursuer and evader satisfy their min and max needs respectively. Simulations show that with the increase of pursuit-evasion area, capture frequency has robust convergence, and average capture time and iterative frequency increase faster.