What is it about?
This article is a comprehensive survey of how deep learning—an advanced type of artificial intelligence—is used in autonomous driving. It summarizes key research developments across perception (e.g., recognizing objects and road signs), localization (knowing where the vehicle is), and decision-making (planning how to drive). We explain various deep learning models like convolutional neural networks, transformers, and reinforcement learning, and show how they contribute to the design of autonomous vehicles. The survey also compares different system architectures, from traditional pipeline-based setups to modern end-to-end AI models. It serves as a one-stop resource for anyone interested in understanding how AI research powers self-driving technologies.
Featured Image
Photo by Remy Gieling on Unsplash
Why is it important?
As autonomous vehicles move from concept to reality, it is crucial to understand the technologies that make them possible. Deep learning plays a foundational role, yet the field is broad and fragmented. This survey consolidates knowledge from hundreds of studies to create a structured, accessible overview of deep learning applications in self-driving. It helps researchers, engineers, and policymakers gain a clearer picture of current progress, limitations, and future directions. By doing so, it supports more informed development of safe and efficient autonomous systems.
Perspectives
This survey was motivated by the rapid pace of research in autonomous driving and the growing need for synthesis. Instead of adding another isolated result, our goal was to connect the dots—helping both newcomers and experts understand how deep learning techniques are being applied, evaluated, and integrated. We believe this work can serve as a valuable reference point and teaching tool, especially for those entering this interdisciplinary field.
Jingyuan Zhao
University of California Davis
Read the Original
This page is a summary of: A Survey of Autonomous Driving from a Deep Learning Perspective, ACM Computing Surveys, April 2025, ACM (Association for Computing Machinery),
DOI: 10.1145/3729420.
You can read the full text:
Contributors
The following have contributed to this page
