A Review of Lane Detection in Autonomous Vehicles

Abstract

With the rapid development of autonomous driving technology, lane detection technology, as a key component, has made significant progress in recent years. This paper reviews the classification and principles of lane detection technology, including methods based on visual, LiDAR, millimeter wave radar, and multi-sensor fusion. In terms of the latest research progress, 3D lane detection methods based on deep learning have become a hot topic, including methods based on CNN and Transformer, as well as innovative single-eye 3D lane detection technology. At the same time, the introduction and improvement of large-scale real-world datasets and the development of evaluation indicators have driven the technology forward. However, 3D lane detection technology still faces challenges such as complex environments, sensor technology limitations, computational resource constraints, and algorithm complexity. In the future, 3D lane detection technology will move towards more intelligent, precise, and efficient development, including further application of deep learning technology, deepening of multi-sensor fusion technology, widespread use of high-precision maps, design of lightweight neural network architectures, and acceleration of standardization and standardization processes.