Optimizing Agent Training with Deep Q-Learning on a Self-Driving Reinforcement Learning Environment Artigo de Conferência