Comparing Different Representations For Reinforcement Learning For Automated Vehicles