Generating Adjacency-Constrained Subgoals in Hierarchical Reinforcement Learning

Goal-conditioned hierarchical reinforcement learning (HRL) is a promising approach for scaling up reinforcement learning (RL) techniques.