A structured review of SLAM generation in the AEC industry: Technical framework, site-specific challenges, and adaptive strategies

Abstract

The construction industry has remained relatively stagnant in productivity due to limited automation adoption compared to other sectors. Simultaneous Localization and Mapping (SLAM) technologies have been increasingly introduced to address the challenges of robot autonomy in dynamic, unstructured construction environments. However, existing reviews predominantly regard SLAM as a supporting component, without systematically examining the generation processes and site-specific challenges. This study conducts a structured review of SLAM generation in Architecture, Engineering, and Construction (AEC) robotics, encompassing sensor inputs, feature extraction, state estimation, backend optimization, and map construction. Bibliometric and content analyses identify key limitations arising from environmental degradation and task-environmental misalignment. Adaptive strategies, including sensor fusion, semantic enrichment, and dynamic map updating, are critically evaluated. Despite notable advancements, significant gaps persist in achieving seamless Building Information Modelling (BIM) integration and robust real-time adaptability. Future research directions are proposed, emphasizing multi-sensor fusion, semantic-aware mapping, collaborative SLAM, and data standardization to support scalable deployment in smart construction contexts.