Abstract
With the rapid development of artificial intelligence, computer vision, and intelligent transportation technologies, autonomous driving has become an important research direction in academia and industry. Deep learning, with its powerful feature extraction, pattern recognition, and temporal modeling capabilities, has shown significant advantages in autonomous driving environmental perception, trajectory prediction, decision planning, and vehicle control. Convolutional neural networks have played a crucial role in object detection and semantic segmentation tasks, effectively improving the understanding of complex road scenarios by autonomous driving systems. Recurrent structures such as Long Short-Term Memory networks have high application value in temporal modeling and trajectory prediction. Transformer models, with their self-attention mechanism, excel in long-distance dependency modeling, multimodal fusion, and high-level decision planning. Furthermore, end-to-end autonomous driving methods, by integrating perception, decision-making, and control into a unified learning framework, provide new ideas for optimizing the overall performance of autonomous driving systems. This paper reviews the key applications of deep learning in autonomous driving, systematically analyzes its research progress in environmental perception, path decision-making, vehicle control, and end-to-end autonomous driving, and summarizes the main challenges and future development trends, aiming to provide a reference for related research.
