Ar-guided modular robotic bricklaying: A proof-of-concept system for human-in-the-loop construction workflows

Purpose – In recent years, research in construction robotics has significantly advanced within the construction sector. However, many collaborative robotic systems have yet to be effectively integrated into large-scale construction settings due to limitations such as restricted workspace, deployment complexity and insufficient adaptability to dynamic on-site conditions. To address this issue, this paper aims to establish an augmented reality (AR)-assisted human-robot collaboration (HRC) workflow that uses low-fidelity visual markers to spatially register a modular robotic unit with a digital wall model, while enabling human-guided robot relocation supported by real-time feasibility visualisation. Design/methodology/approach – The authors evaluate the complete cyber-physical workflow by constructing two large-scale masonry-like walls, each spanning 3 m, exceeding the operational range of the collaborative robotic arm and requiring repeated manual relocation of the modular robotic unit. After assembly, both structures were 3D-scanned and compared with their 3D digital models using cloud-to-cloud (C2C) deviation analysis. Segment-level deviation was used to assess the reliability of robotic pick-and-place execution, while global deviation quantified spatial drift introduced by the AR-based registration during repeated relocation cycles. Findings – Cluster deviation analysis confirms consistent robotic pick-and-place accuracy across repeated assembly cycles, demonstrating reliable task-level execution. However, global deviation results reveal accumulated spatial drift caused by AR-based registration, indicating that the proposed workflow is most suitable for construction scenarios requiring moderate accuracy (˜ ±5 mm tolerance) rather than high-precision industrial fabrication. Research limitations/implications – The study's accuracy was limited by the use of only two visual markers, increasing susceptibility to AR spatial drift during repeated registration. The geometric simplicity of the dry-stacked walls restricted evaluation of more complex construction scenarios, and the controlled laboratory lighting conditions did not reflect the variability of real construction environments. In addition, the robotic arm's operational range was constrained by the geometry of the mobile platform. Future work should investigate the use of additional markers or sensor-fusion methods, evaluate performance in real on-site conditions, and explore improved mobile platform designs to extend reachability and robustness. Practical implications – The study demonstrates that AR-assisted spatial anchoring can improve the accessibility of robotic bricklaying in construction by reducing reliance on complex simultaneous localisation and mapping-based localisation and costly autonomous platforms. The proposed workflow allows operators to manually reposition modular robotic units using intuitive AR feedback, enabling reliable assembly without advanced robotics expertise. This approach lowers implementation complexity, reduces training requirements and supports wider adoption of collaborative robotic systems in small-to-medium construction projects. Social implications – The study supports a more human-centred model of construction automation by positioning robots as collaborative tools rather than replacements for labour. By simplifying robot operation through AR guidance, the proposed workflow can reduce workforce resistance often associated with job displacement and technological complexity. The system allows operators to retain users' agency through manual robot repositioning while benefiting from robotic precision during execution. This approach may facilitate workforce upskilling, reduce physical strain and improve safety, while broadening access to robotic technologies across diverse construction teams, thereby supporting more inclusive and socially sustainable adoption of automation in the construction sector. Originality/value – This study contributes (1) a validated human-in-the-loop workflow for large-scale robotic b icklaying that combines human-guided robot relocation with robot-executed geometric placement, (2) an AR-based spatial anchoring method using low-fidelity visual markers to support repeated robot re-registration during large-scale assembly, and (3) an AR-supported inverse-kinematics feasibility visualisation that enables informed relocation decisions across repeated construction cycles.