The Use of Domain Knowledge in Reinforcement Learning door