One embodiment is directed to an approach to solving a leader-follower problem in which a leader has a set of leader actions and a follower has a set of follower actions. The approach includes receiving an expression of the leader-follower problem as a normal form Stackelberg game. The approach further includes, for each possible follower action, solving a linear program (LP) problem to determine a respective optimal mixed leader strategy, wherein the LP problem optimizes a leader payoff for a given mixed leader strategy and a given fixed follower action over a feasible region that includes only mixed leader strategies that provoke that respective follower action. The approach further includes generating an output derived from the optimal mixed leader strategies, and outputting the output by changing a physical state associated with an interface.