An Integrated Mobile Robot Framework for Worker Safety, Defect, and Thermal Monitoring in Construction

Accurate sensing is essential for automated inspection in complex construction environments. This study develops and field-validates an integrated mobile-robot framework combining worker localization, structural crack detection, and thermal monitoring within a shared spatial reference frame. A mobile robot equipped with LiDAR, RGB, and thermal cameras integrates ArUco-based worker localization, YOLOv5-LiDAR fusion for crack detection and extent estimation, and synchronized thermal-RGB monitoring. Field tests in a newly completed multi-story building showed a worker-localization accuracy of 81–95% (mean 90%; mean positioning error 1.09 m, SD 0.83 m, across six valid trials) and up to 96% and 99% accuracy for crack extent (bounding-box span) and location estimation, respectively. The thermal module successfully captured spatially referenced thermal-RGB image pairs. Unlike studies validating individual sensing modules, this work demonstrates an integrated multi-sensor robotic inspection framework under real building conditions and provides practical guidance for construction management. Given the use of printed crack surrogates, the absence of genuine thermal anomalies during testing, and the lack of baseline comparisons, the reported figures should be read as validating the sensing-and-mapping pipeline and workflow—not as field-proven defect-detection accuracy—with real-defect validation, controlled anomaly induction, and baseline benchmarking identified as the necessary next steps.