Reproduce “Building Information Models to Robot-Ready Site Digital Twins (BIM2RDT): An Agentic AI Safety-First Framework” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Reza Akhavian, Mani Amani, Johannes Mootz, Robert Ashe, Behrad Beheshti
Researchers at San Diego State University and UC San Diego developed BIM2RDT, an agentic AI framework that converts static BIM into dynamic, robot-ready digital twins for construction sites, emphasizing safety. The system features a Semantic-Gravity ICP (SG-ICP) algorithm, which improves point cloud registration accuracy by up to 88.3% RMSE and integrates real-time worker safety monitoring for proactive interventions.
Key insights
- Integrating semantic information from BIM and LLMs can provide object-adaptive priors that significantly improve the robustness and physical plausibility of point cloud registration.
- An agentic AI approach facilitates a continuous feedback loop where robot-collected data updates the digital twin, which in turn optimizes subsequent robot missions and safety management decisions.
- Real-time worker safety monitoring, such as Hand-Arm Vibration exposure, can be effectively mapped to digital twin entities using IFC standards, enabling automated, proactive safety interventions.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.