Abstract: In recent years, academic research in the field of autonomous driving (AD) has gained significant attention. Intelligent driving is a multi-disciplinary field, and the application of deep reinforcement learning (DRL) to driving strategies, control, and related areas has yielded notable results. We provide a comprehensive review of this body of work from three perspectives: First, we introduce the basic concepts, mathematical modeling and algorithm classification of reinforcement learning. Second, we comprehensively review the applications of DRL in autonomous driving decision-making and control. Finally, we conduct an in-depth discussion on how DRL models address critical challenges in AD applications, such as driving safety, interaction with other traffic participants, and sample efficiency, and we offer perspectives on future research directions in this field.