This survey provides an overview of deep reinforcement learning (DRL) and its application in autonomous driving. It discusses the key challenges and opportunities for applying DRL in real-world autonomous driving systems, as well as related domains such as behavior cloning, imitation learning, and inverse reinforcement learning. The paper also addresses the role of simulators in training agents, methods for validating and robustifying solutions, and the computational challenges in deploying DRL in autonomous driving. The paper outlines the components of an autonomous driving system, including perception, localization and mapping, planning and driving policy, and control. It discusses the use of reinforcement learning (RL) in autonomous driving, including the different types of RL algorithms such as value-based, policy-based, and actor-critic methods. The paper also covers the differences between model-based and model-free approaches, as well as on-policy and off-policy methods. The paper reviews the main extensions to RL, including reward shaping, multi-agent reinforcement learning, multi-objective reinforcement learning, and state representation learning. It also discusses learning from demonstrations, including behavior cloning and inverse reinforcement learning, and how these methods can be combined with reinforcement learning to improve performance. The paper highlights the importance of simulators in training agents and the challenges of deploying RL in real-world autonomous driving systems. It also discusses the computational challenges and risks associated with applying current RL algorithms, such as imitation learning and deep Q-learning, in autonomous driving. The paper concludes by emphasizing the need for further research in this area and the potential of DRL in autonomous driving.This survey provides an overview of deep reinforcement learning (DRL) and its application in autonomous driving. It discusses the key challenges and opportunities for applying DRL in real-world autonomous driving systems, as well as related domains such as behavior cloning, imitation learning, and inverse reinforcement learning. The paper also addresses the role of simulators in training agents, methods for validating and robustifying solutions, and the computational challenges in deploying DRL in autonomous driving. The paper outlines the components of an autonomous driving system, including perception, localization and mapping, planning and driving policy, and control. It discusses the use of reinforcement learning (RL) in autonomous driving, including the different types of RL algorithms such as value-based, policy-based, and actor-critic methods. The paper also covers the differences between model-based and model-free approaches, as well as on-policy and off-policy methods. The paper reviews the main extensions to RL, including reward shaping, multi-agent reinforcement learning, multi-objective reinforcement learning, and state representation learning. It also discusses learning from demonstrations, including behavior cloning and inverse reinforcement learning, and how these methods can be combined with reinforcement learning to improve performance. The paper highlights the importance of simulators in training agents and the challenges of deploying RL in real-world autonomous driving systems. It also discusses the computational challenges and risks associated with applying current RL algorithms, such as imitation learning and deep Q-learning, in autonomous driving. The paper concludes by emphasizing the need for further research in this area and the potential of DRL in autonomous driving.