Intersection Information Modelling and Case Analysis under the Dual Drive of Path Guidance and Scenario Constraints

This study proposes a novel intersection-centric information framework designed to provide a new perspective and practical approach for enabling autonomous vehicles (AVs) to navigate intersections safely and efficiently. While high-definition (HD) maps provide rich spatial information for autonomous driving, their excessive data volume often leads to increased computational burden and delays, particularly in intersection environments where timely positioning, perception, and decision-making are critical. To address this issue, we introduce the Road-Lane-Device-Scenario (RLDS) model, which separates and extracts key static and dynamic intersection elements, focusing on essential semantic features such as lane topology, traffic signal logic, and right-of-way rules. This targeted framework introduces an innovative approach to reduce data redundancy, thereby providing significant potential for enhancing the efficiency and intelligence of information processing and decision-making in autonomous driving systems. By effectively addressing the critical bottlenecks associated with intersection scenarios, the proposed framework demonstrates strong potential for practical application, enabling safer, more reliable, and more efficient autonomous vehicle operations in complex urban environments.