SGR-BIM: New AI framework automates geometry compliance checks with 84.3% accuracy
Graph-based reasoning boosts fire safety compliance by 8.6% over single-agent baselines.
A team of researchers (Zixuan Xiao, Pei Troh Koh, Jun Ma, Jack C.P. Cheng) has proposed SGR-BIM, a graph-driven reasoning framework that tackles the longstanding challenge of automating geometry-intensive compliance checks in Building Information Modeling (BIM). Current methods rely on static rule templates that fail to handle multi-hop reasoning chains or latent spatial dependencies across multiple building entities. SGR-BIM overcomes this by constructing a cross-modal knowledge graph that dynamically integrates user intent, regulatory semantics, and BIM geometry, enabling interpretable reasoning without rigid hard-coding.
Validated on 679 expert-verified queries from fire safety codes, the framework achieves 84.3% accuracy—an 8.6% improvement over enhanced single-agent baselines. The research, published in Automation in Construction and arXiv (2606.12065), represents a significant step toward transparent, flexible automated compliance workflows in the Architecture, Engineering, and Construction (AEC) industry. By making the reasoning process interpretable and adaptable, SGR-BIM could reduce manual review time and errors in regulatory checks for complex building geometries.
- SGR-BIM uses a cross-modal knowledge graph to align user intent, regulatory semantics, and BIM geometry.
- Achieves 84.3% accuracy on 679 fire safety queries, 8.6% better than enhanced single-agent baselines.
- Framework avoids rigid hard-coding, enabling interpretable multi-hop reasoning for complex spatial dependencies.
Why It Matters
This graph-based reasoning paradigm could automate tedious, error-prone geometry checks in building compliance, saving time and improving safety.