Dr. Long Chen

City University of Hong Kong

Talk Title

Human-AI Collaborative Decision-making for Building Asset and Fire Safety Management

Abstract

Managing building assets and fire safety throughout their lifecycle necessitates a seamless integration of domain expertise and computational intelligence. This research presents a comprehensive programme on Human-AI Collaborative Decision-making, introducing a methodological pipeline that addresses four critical and interconnected stages: design compliance, construction-to-operation model updating, in-service inspection, and long-term asset management. Our pipeline initiates at the design stage, where an automated system verifies layout compliance against fire codes and employs a connectivity-guided diffusion model to generate qualified redesigns, transforming a traditionally manual, hours-long expert process into a convenient, iterative solution. Post-construction, our research introduces two complementary methods for updating digital models to reflect as-built conditions: (1) an adaptive 3D Gaussian Splatting technique that achieves a 57% reduction in processing time for change integration, and (2) a semi-automated BIM workflow that reduces manual effort by ~70% for piping system updates. For operational safety, a three-layer knowledge-based system automates compliance checking. It integrates computer vision and a specialised Visual Question Answering (VQA) module to extract evidence from imagery, fuses it within an extended Brick ontology, and executes regulatory rules via SPARQL to deliver traceable compliance decisions. Finally, all geometric, semantic, and compliance records are consolidated into an IFC-based asset management platform enhanced with Bayesian data fusion, enabling continuous risk assessment and lifecycle-oriented decision support. This research demonstrates and can benefit a paradigm shift towards proactive, evidence-based, and collaborative lifecycle management of built environments, where human expertise is augmented by a continuous stream of AI-processed insights.

Biography

Prof. Long Chen is Assistant Professor at Department of Architecture and Civil Engineering, City University of Hong Kong. Before joining CityUHK, he worked as Senior Lecturer/Lecturer in Surveying at Loughborough University, Co-Director of LU-WTW Mini-CDT, and also affiliated as the Visiting Researcher at the Alan Turing Institute (ATI). He also used to work as a Research Associate at Imperial College London, and as a Data Scientist at ATI. He obtained a BEng (Hons) in Hydraulic Engineering at Tsinghua University in 2014, and a PhD in Civil Engineering from the University of Hong Kong in 2019. His research expertise encompasses development and deployment of advanced computing techniques, such as AI, computer vision, and Large Language Models (LLMs), for object recognition and monitoring, digital twinning complex built environments and infrastructures, and operation management of built assets. Prof CHEN has secured over HK$10 million of research grants focused on AI and digital technologies from Hong Kong RGC, Royal Society, Innovate UK, ICE and industrial funds. He has published 2 books and more than 60 peer-reviewed scholarly items covering topics on AI, computer vision, imagery processing, point clouds, and digital twins etc. Prof Long Chen has won the ICE Charles Manby Prize (2024), Loughborough University Early-career Achievement Award (2022) and Outstanding Performance Award (2023), and three Best Paper/Presentation Awards etc.