Building Success: Generative AI-Powered Multimodal Construction Issue Identification and Knowledge-Driven Mitigation

Generative AI is redefining the future of construction by enabling seamless integration of multimodal data to address long-standing industry challenges. Our research focuses on leveraging these capabilities to uncover and mitigate major project issues. Key efforts include extracting risk insights from historical claims and supplemental agreements using BERTopic modeling, applying vision-language models to detect hazardous materials, and developing the Meet2Mitigate framework to automatically capture and summarize key issues from dynamic construction meetings. In parallel, we are curating a comprehensive multimodal knowledge base consisting of best practices, lessons learned, and educational resources. This knowledge base is integrated with multimodal retrieval-augmented generation (RAG) to provide targeted mitigation strategies that directly support the resolution of project-specific issues.

Building Harmony: Human-Centric Edge Intelligence Design for Seamless Human–Technology Interaction on Construction Sites

The central thrust of this research lies in embedding artificial intelligence (AI) within edge computing environments to enable the widespread deployment of edgeAI solutions on construction sites, where network connectivity and bandwidth are often constrained. Our current efforts focus on integrating advanced computer vision models into human-centered edge devices to support real-time material classification, hazard detection, and object recognition. Looking ahead, the rapid emergence of large language models (LLMs) and vision-language models (VLMs) presents exciting opportunities to enhance edge intelligence. Future work will explore the integration of these multimodal models into edge systems and develop intuitive, worker-friendly interfaces that empower construction personnel to seamlessly interact with generative AI applications in physical site environments. The ultimate goal is to harness the capabilities of edgeAI to drive the adoption of advanced technologies and accelerate the evolution of construction automation.

Building Trust: Blockchain-Enabled Smart Contracts for Decentralized and Autonomous Construction Management

Blockchain-enabled smart contracts represent a tremendous opportunity for enabling decentralized construction management by embedding predefined contractual clauses in smart contracts and integrating external verification mechanisms. Our research first focuses on formulating smart contract clauses to coordinate task assignments and payment distribution among subcontractors. To achieve this, a simulation model was employed to generate various “what-if” construction scenarios. Then, game theory principles were applied to form the equitable reward-penalty sharing contract clauses under different scenarios, ensuring fair compensation structures aligned with subcontractor performance. The second focus involves automating smart contract execution through the integration of as-built images and computer vision models, which function as blockchain oracles to assess project progress in real-time. Collectively, our work advances the development of a decentralized autonomous organization (DAO) capable of managing construction activities with minimal centralized governance. Centralized project managers can delegate specific tasks to the system for automated execution.