Temporal-Logic-Based Reward Shaping for Continuing Learning Tasks