Spatial-temporal traceability and coordination in modular integrated construction: A future framework with hybrid digital twin and reinforcement learning

Spatial-temporal traceability and coordination in Modular Integrated Construction (MiC) play a pivotal role in achieving seamless integration of lifecycle and effective management of heterogeneous entities. The significant enabler lies in multi-modal data acquisition and fusion, cyberphysical traceability, and coordinated decision-making. However, outdated information and automation technologies in MiC represent a significant barrier for the next-generation upgrade. Digital Twin (DT) and Reinforcement Learning (RL) have the potential complementarity as the state-of-the-art technological fusion to offer data-driven traceability and coordinated decisionmaking. This paper conducts a systematic scientometric review through bibliometric selection and scientometric analysis for the representative literature on DT and RL-enabled MiC. 108 highquality articles are selected through academic database searches, well-defined search strings, and rigorous inclusion and exclusion criteria. The key challenges are identified and systematically categorized into four critical layers of DT and four core limitations of RL. A lifecycle management framework is proposed, including the controlled and programmable entities, graph-based hierarchical data management, system-of-system-level virtual modeling, multiagent decision-making, and lifecycle services in MiC. This review offers practical guidance for MiC practitioners on how to adopt a cross-disciplinary technological integration to enhance spatialtemporal traceability and coordination.