Research on the Strategy for Establishing National 3D Spatial Information Infrastructure to Support Physical AI in the AX Era; [AI 시대의 Physical AI 지원을 위한 국가 3차원 공간정보 인프라 구축 전략 연구]

초록

This study presents a strategy for establishing national 3D spatial information infrastructure to support Physical AI, such as autonomous driving and UAM (Urban Air Mobility), in the era of AX (AI Transformation). Existing 2D digital maps and visualization-oriented 3D data lack the precision and semantic information necessary for direct machine interpretation and computation, leading to a "Sim-to-Real Gap" that hinders technical advancement. To address these limitations, this research defines six core datasets—Point Cloud (min. 50 pts/㎡), Buildings (CityGML 3.0 LoD 2), Terrain (nationwide 0.5m grid), Roads, Railways, and Rivers—as an intelligent "World Model" for Physical AI. This paper proposes the adoption of hybrid sensor technology, which simultaneously acquires aerial imagery and LiDAR data, along with AI-based automation pipelines to ensure economic efficiency and real-time updates of nationwide data. Furthermore, it elaborates on the necessity of semantic ontology modeling to bridge the gap between geometric representation and logical reasoning for AI agents. The strategies outlined in this study are expected to enhance national disaster management capabilities through integration with water management and forest digital twin platforms, as well as providing essential data for K-UAM route planning and vertiport site analysis. Ultimately, a machine-readable 3D spatial information infrastructure will serve as a critical asset for South Korea to lead global AX industry standards and secure digital spatial sovereignty. © 2026, Korean Society of Surveying. All rights reserved.

키워드