A Review of Environmental Perception Technology Based on Multi-Sensor Information Fusion in Autonomous Driving

Environmental perception is a key technology for autonomous driving, enabling vehicles to analyze and interpret their surroundings in real time to ensure safe navigation and decision-making. Multi-sensor information fusion, which integrates data from different sensors, has become an important approach to overcome the limitations of individual sensors. Each sensor has unique advantages. However, its own limitations, such as sensitivity to lighting, weather, and range, require fusion methods to provide a more comprehensive and accurate understanding of the environment. This paper describes multi-sensor information fusion techniques for autonomous driving environmental perception. Various fusion levels, including data-level, feature-level, and decision-level fusion, are explored, highlighting how these methods can improve the accuracy and reliability of perception tasks such as object detection, tracking, localization, and scene segmentation. In addition, this paper explores the critical role of sensor calibration, focusing on methods to align data in a unified reference frame to improve fusion results. Finally, this paper discusses recent advances, especially the application of machine learning in sensor fusion, and highlights the challenges and future research directions required to further enhance the environmental perception of autonomous systems. This study provides a comprehensive review of multi-sensor fusion technology and deeply analyzes the advantages and challenges of different fusion methods, providing a valuable reference and guidance for the field of autonomous driving.