Enhancing Autonomy And Efficiency In Goal-Conditioned Reinforcement Learning