A Framework for Scalable LLM Applications in Robotics-Enabled Construction Management

Abstract

The integration of large language models (LLMs), artificial intelligence, and robotics is creating new opportunities for improving construction management, particularly in environments characterized by complex scheduling, distributed information, workforce coordination, and changing operational conditions. However, the practical deployment of LLM applications in robotics-enabled construction remains constrained by scalability, interoperability, contextual reliability, and the need to connect language-based reasoning with physical construction processes. This research develops a conceptual framework for scalable LLM applications in robotics-enabled construction management by synthesizing the provided literature on vocational education, organizational transformation, knowledge systems, and digitally supported decision processes. The framework conceptualizes LLMs as an intelligent coordination layer between construction data, human decision-makers, and robotic systems. It emphasizes five interconnected functions: knowledge interpretation, schedule reasoning, task allocation, human–robot communication, and adaptive decision support. The framework further considers scalability through modular architecture, data interoperability, human oversight, and continuous feedback. The analysis indicates that LLM-enabled construction management can potentially improve coordination and responsiveness when language intelligence is connected to structured operational data and robotic execution mechanisms. Nevertheless, the framework also identifies important limitations involving hallucination, data quality, domain adaptation, explainability, and human accountability. The study contributes a research-oriented architecture for examining how scalable LLM applications can be integrated into construction management while preserving operational control and human-centered decision-making.