An Automated Pipeline for Generating a Multi-Layered Spatial Map for Indoor Robot Navigation from BIM Data

초록

Accurate and machine-readable indoor maps are essential for autonomous systems, such as mobile robots, to operate effectively. Existing traditional mapping methods, such as manual piloting or Simultaneous Localization and Mapping (SLAM), are often time-consuming, lack semantic details, and are affected by environmental changes. Building Information Modeling (BIM) has recently been utilized to provide detailed building data, but it is limited in its direct deployment for robotics applications. To address these limitations, this paper proposes a fully automated pipeline and a standardized, multi-layer map data model that converts Building Information Models (BIM), specifically in the Industry Foundation Classes (IFC) format, into robot-usable navigation maps. The proposed technique generates an integrated map model comprising three distinct yet interconnected layers. These layers include (1) a metric occupancy grid, (2) a semantic entity layer, and (3) a topological graph. The generated Spatial Map Model was evaluated in a simulation experiment to verify its application. The findings illustrated the pipeline’s robustness in automatically producing an integrated, multi-layer map that improves localization and enables context-aware navigation for robots in an indoor environment.

키워드