This paper surveys the current state-of-the-art in deep learning technologies used in autonomous driving. It begins by introducing AI-based self-driving architectures, including convolutional and recurrent neural networks, as well as deep reinforcement learning. These methodologies form the basis for driving scene perception, path planning, behavior arbitration, and motion control. The paper investigates both modular perception-planning-action pipelines and end-to-end systems that directly map sensory information to steering commands. It also addresses current challenges in designing AI architectures for autonomous driving, such as safety, training data sources, and computational hardware. The survey provides insights into the strengths and limitations of deep learning and AI approaches for autonomous driving and assists with design choices. The paper discusses various deep learning technologies used in autonomous driving, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Deep Reinforcement Learning (DRL). CNNs are used for processing spatial information, such as images, and can automatically learn feature representations. RNNs are particularly effective for processing temporal sequence data, such as text or video streams. DRL is used for learning optimal driving policies in autonomous driving tasks. The paper also covers deep learning for driving scene perception and localization, including object detection, semantic and instance segmentation, and localization. It discusses the use of LiDAR and camera-based sensing technologies, as well as the advantages and disadvantages of each. The paper also explores the use of occupancy maps for perception and the challenges of using deep learning for path planning and behavior arbitration. Finally, the paper discusses motion controllers for AI-based self-driving cars, including learning controllers and end-to-end control systems. It highlights the importance of safety, training data, and computational hardware in the design of autonomous driving systems. The paper concludes with a discussion of the future directions for deep learning in autonomous driving.This paper surveys the current state-of-the-art in deep learning technologies used in autonomous driving. It begins by introducing AI-based self-driving architectures, including convolutional and recurrent neural networks, as well as deep reinforcement learning. These methodologies form the basis for driving scene perception, path planning, behavior arbitration, and motion control. The paper investigates both modular perception-planning-action pipelines and end-to-end systems that directly map sensory information to steering commands. It also addresses current challenges in designing AI architectures for autonomous driving, such as safety, training data sources, and computational hardware. The survey provides insights into the strengths and limitations of deep learning and AI approaches for autonomous driving and assists with design choices. The paper discusses various deep learning technologies used in autonomous driving, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Deep Reinforcement Learning (DRL). CNNs are used for processing spatial information, such as images, and can automatically learn feature representations. RNNs are particularly effective for processing temporal sequence data, such as text or video streams. DRL is used for learning optimal driving policies in autonomous driving tasks. The paper also covers deep learning for driving scene perception and localization, including object detection, semantic and instance segmentation, and localization. It discusses the use of LiDAR and camera-based sensing technologies, as well as the advantages and disadvantages of each. The paper also explores the use of occupancy maps for perception and the challenges of using deep learning for path planning and behavior arbitration. Finally, the paper discusses motion controllers for AI-based self-driving cars, including learning controllers and end-to-end control systems. It highlights the importance of safety, training data, and computational hardware in the design of autonomous driving systems. The paper concludes with a discussion of the future directions for deep learning in autonomous driving.