4D BIM-Driven Framework for Automated Construction Progress Monitoring Using MCP-Enabled LLM

Effective construction progress monitoring is crucial for mitigating project delays, yet conventional methods remain labor-intensive. This study proposes an automated monitoring framework leveraging a Large Language Model (LLM) equipped with the Model Context Protocol (MCP) to orchestrate the workflow from image analysis to 4D Building Information Model (BIM) updates. The framework employs an LLM orchestrator to autonomously execute four stages: (1) matching site images with 4D BIM elements, (2) generative AI-based 3D reconstruction and scale calibration, (3) volume-based progress assessment, and (4) automated updates to the as-built 4D BIM. Validated via a masonry wall mock-up using miniature bricks, the case study successfully reconstructed a partially completed wall from a single image. It calculated a Completion Ratio (CR) of 91.4%, classified the status as 'DELAYED', and automatically updated the 4D BIM. These results demonstrate that MCP-enabled generative AI can effectively automate monitoring and enhance decision-making by reflecting actual site conditions in 4D BIM.