A reward shaping technique that adds auxiliary rewards based on a potential function while guaranteeing the optimal policy remains unchanged.