Finding the optimal strategies for robotic patrolling with adversaries in topologically-represented environments