Tag
This paper introduces G2QDR, a framework that leverages directed state graphs to generate dense rewards for goal-conditioned hierarchical reinforcement learning, enhancing performance in sparse reward environments.