pith. sign in

arxiv: 2605.20264 · v1 · pith:XJEBPD4Znew · submitted 2026-05-18 · 💻 cs.RO · cs.HC

Adaptive Human-Robot Collaboration for Masonry Construction Under Material and Assembly Uncertainty

Jutang Gao (1) , Arash Adel (1) ((1) Princeton University) This is my paper

Pith reviewed 2026-05-21 07:36 UTC · model grok-4.3

classification 💻 cs.RO cs.HC
keywords human-robot collaborationmasonry constructionspatial projectionlaser scanningtolerance accumulationadaptive workflowbricklayingconstruction robotics
0
0 comments X

The pith

An end-effector-mounted projector and laser scanning let a robot and human collaborate on brickwork while correcting for material variations and tolerance buildup.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents a workflow for masonry construction in which a robot places bricks while a human applies adhesive, guided by real-time spatial projections from a projector mounted on the robot's end-effector. Laser scans then provide feedback to adjust the robot's grasping and placement poses in response to size differences and accumulated errors. Experiments on full-scale walls in both standard running-bond patterns and irregular layouts showed that the projections improved adhesive consistency and speed while the scans kept courses level and prevented collisions that occur in open-loop operation. A sympathetic reader cares because the approach shows how sensing and projection can make human-robot teams more reliable on site without demanding perfectly uniform materials or flawless pre-planning.

Core claim

The central claim is that mounting a projector on the robot end-effector to deliver spatially registered just-in-time guidance for manual adhesive application, combined with laser scanning for feedback-driven grasping and placement pose correction, enables the collaborative system to adjust human and robotic actions to material variability and accumulated assembly tolerances, thereby improving adhesive consistency, reducing application time, maintaining level courses, and avoiding collision failures in both conventional and nonstandard brick configurations.

What carries the argument

End-effector-mounted projector for spatially registered projection guidance together with laser scanning for real-time feedback-driven pose correction.

Load-bearing premise

The projector must deliver accurate, visible, and usable guidance for adhesive application under real construction lighting and surface conditions, and laser scanning must supply timely and precise enough feedback to correct errors before tolerances cause failures.

What would settle it

A demonstration in which bright site lighting renders the projector guidance unusable or laser scans arrive too late to prevent level deviations and collisions when brick dimensions vary beyond the tested range.

Figures

Figures reproduced from arXiv: 2605.20264 by Arash Adel (1) ((1) Princeton University), Jutang Gao (1).

Figure 1
Figure 1. Figure 1: Workcell setup for augmented human–robot collaborative masonry construction. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Collaborative masonry workflow integrat [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: A full-frame rectangle is cast onto the planar sur [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: The experiment without brick pose correc [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Variations in brick configuration, includ [PITH_FULL_IMAGE:figures/full_fig_p005_6.png] view at source ↗
Figure 4
Figure 4. Figure 4: Brick pose correction based on brick-level [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 7
Figure 7. Figure 7: Documented experiment processes: a. robotic manipulator and a finished nonstandard construct; b. [PITH_FULL_IMAGE:figures/full_fig_p006_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: Height mean absolute error (MAE) analy￾sis based on course-level scanning results. Current system includes limited integration of active safety monitoring and a constrained human-to-robot com￾munication interface, which currently relies on indirect terminal-based input. Future work will expand the system in several directions: incorporating direct communication modalities, such as gesture and speech, to su… view at source ↗
Figure 8
Figure 8. Figure 8: Adhesive coverage identification and anal [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
read the original abstract

Human-robot collaboration in construction is often challenged by limited robot-to-human communication and the need to adapt to tolerance accumulation arising from material and assembly uncertainties. We present an adaptive human-robot collaborative workflow for masonry construction that addresses communication limitations and tolerance accumulation, demonstrated through a brickwork case study in which a robot places bricks while a human applies adhesive. This workflow is enabled by two complementary mechanisms: 1) an end-effector-mounted projector that provides spatially registered, just-in-time projection guidance for manual adhesive application, and 2) laser scanning for feedback-driven grasping and placement pose correction. Together, these mechanisms enable adjustment of human and robotic actions in response to material variability and accumulated assembly tolerances. Full-scale experiments across conventional running-bond and nonstandard configurations demonstrate that projection guidance improves adhesive application consistency and reduces application time, while laser-based correction maintains level courses and avoids collision-prone failures associated with open-loop execution. These results indicate that integrating spatial projection with feedback-driven adaptation, enabled by material and as-built sensing, can mitigate tolerance accumulation and improve precision and robustness in human-robot collaborative construction.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper claims that integrating an end-effector-mounted projector for spatially registered adhesive guidance with laser-scanning-based feedback for grasping and placement correction enables adaptive human-robot collaboration in masonry construction, mitigating tolerance accumulation from material and assembly uncertainties. This is demonstrated via full-scale experiments on running-bond and nonstandard brick configurations, where projection improves adhesive consistency and reduces time while laser correction maintains level courses and avoids open-loop failures.

Significance. If the experimental results hold under realistic site conditions, the work would represent a meaningful advance in human-robot collaboration for construction by providing concrete mechanisms to address communication gaps and tolerance buildup, with potential applicability to other assembly tasks involving uncertainty. The physical demonstrations on both standard and nonstandard configurations strengthen the practical relevance.

major comments (2)
  1. [Abstract] Abstract: The central claim that the integrated projection and laser feedback 'mitigate tolerance accumulation and improve precision and robustness' is load-bearing but rests on reported positive outcomes without any quantitative metrics, error bars, statistical analysis, or details on data collection/exclusion criteria for the full-scale experiments. This makes it difficult to verify the magnitude of improvement or rule out confounds.
  2. [Results/Experiments] Experimental validation (implied in results section): The weakest link identified is the assumption that the end-effector projector delivers visible, usable guidance under variable construction lighting and that laser scans provide timely corrections before tolerance propagates; the manuscript provides no measured data on projector contrast, scan-to-correction latency, or residual tolerance per course to substantiate that these mechanisms function as asserted in real conditions.
minor comments (2)
  1. [Methods] Clarify the exact sensing hardware, registration method for the projector, and update rate of the laser feedback loop to allow replication.
  2. Add a limitations section discussing performance under varying illumination, dust, or time constraints typical of construction sites.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive review and recognition of the practical relevance of our adaptive human-robot collaboration approach. We address each major comment below and have revised the manuscript to provide stronger quantitative support for our claims.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that the integrated projection and laser feedback 'mitigate tolerance accumulation and improve precision and robustness' is load-bearing but rests on reported positive outcomes without any quantitative metrics, error bars, statistical analysis, or details on data collection/exclusion criteria for the full-scale experiments. This makes it difficult to verify the magnitude of improvement or rule out confounds.

    Authors: We agree that the abstract would benefit from explicit quantitative support. The full manuscript reports concrete outcomes from the full-scale experiments, including measured reductions in adhesive application time and improved consistency, as well as maintained course levelness that avoided open-loop failures. We have revised the abstract to include these key quantitative metrics (e.g., time reductions and placement precision) and expanded the experimental protocol description in the methods section to detail data collection procedures and inclusion criteria. revision: yes

  2. Referee: [Results/Experiments] Experimental validation (implied in results section): The weakest link identified is the assumption that the end-effector projector delivers visible, usable guidance under variable construction lighting and that laser scans provide timely corrections before tolerance propagates; the manuscript provides no measured data on projector contrast, scan-to-correction latency, or residual tolerance per course to substantiate that these mechanisms function as asserted in real conditions.

    Authors: We concur that direct measurements would strengthen substantiation under realistic conditions. The revised manuscript now includes additional experimental data quantifying projector contrast and visibility under variable construction-site lighting, measured scan-to-correction latencies, and residual placement tolerances per course. These additions confirm that the projection remains usable and that corrections occur in time to prevent tolerance propagation, as evidenced by the maintained level courses in both standard and nonstandard configurations. revision: yes

Circularity Check

0 steps flagged

Empirical experimental workflow with no derivation chain

full rationale

The paper describes an adaptive human-robot masonry workflow enabled by an end-effector projector and laser scanning, validated through full-scale physical experiments on running-bond and nonstandard configurations. Claims of improved adhesive consistency, maintained level courses, and avoided failures rest directly on reported experimental outcomes rather than any equations, fitted parameters, predictions, or first-principles derivations. No load-bearing steps reduce by construction to inputs, self-citations, or ansatzes; the work is self-contained as an empirical demonstration of integrated sensing and adaptation mechanisms.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based on abstract only; no explicit free parameters, axioms, or invented entities are described. The approach relies on standard assumptions in robotics sensing and human-robot interaction.

pith-pipeline@v0.9.0 · 5725 in / 1034 out tokens · 41473 ms · 2026-05-21T07:36:00.740695+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

  1. [1]

    Feedback-Driven Adaptive Multi-Robot Timber Construction.Automation in Construction, 164:105444, 2024

    Arash Adel, Daniel Ruan, Wesley McGee, and Salma Mozaffari. Feedback-Driven Adaptive Multi-Robot Timber Construction.Automation in Construction, 164:105444, 2024

    work page 2024

  2. [2]

    Kamat, and Carol C

    Ci-Jyun Liang, Xi Wang, Vineet R. Kamat, and Carol C. Menassa. Human–Robot Collaboration in Construction: Classification and Research Trends. Journal of Construction Engineering and Manage- ment, 147(10):03121006, 2021. 7

    work page 2021

  3. [3]

    Co-Robotic Assembly of Nonstandard Timber Structures

    Arash Adel. Co-Robotic Assembly of Nonstandard Timber Structures. InHybrids & Haecceities, Pro- ceedings of the 42nd Annual Conference of the Asso- ciation for Computer Aided Design in Architecture (ACADIA), pages 604–613. CumInCAD, 2022

    work page 2022

  4. [4]

    Engrained Perfor- mance: Performance-Driven Computational Design of a Robotically Assembled Shingle Facade System

    Ryan Craney and Arash Adel. Engrained Perfor- mance: Performance-Driven Computational Design of a Robotically Assembled Shingle Facade System. InDistributed Proximities, Proceedings of the 40th Annual Conference of the Association of Computer Aided Design in Architecture (ACADIA), pages 604–

    work page

  5. [5]

    Robotics and Automated Systems in Construction

    Delgado Davila, Manuel Juan, Lukumon Oyedele, Anuoluwapo Ajayi, Lukman Akanbi, Olugbenga Akinade, Muhammad Bilal, and Hakeem Owolabi. Robotics and Automated Systems in Construction. Journal of Building Engineering, 26:100868, 2019

    work page 2019

  6. [6]

    Advancing Robotic Assembly in Construction: Innovations, Challenges, and Opportunities.Automation in Construction, 178: 106370, 2025

    Zhengyi Chen and Arash Adel. Advancing Robotic Assembly in Construction: Innovations, Challenges, and Opportunities.Automation in Construction, 178: 106370, 2025

    work page 2025

  7. [7]

    Reduc- ing Uncertainty in Multi-Robot Construction through Perception Modelling and Adaptive Fabrication

    Daniel Ruan, Wes McGee, and Arash Adel. Reduc- ing Uncertainty in Multi-Robot Construction through Perception Modelling and Adaptive Fabrication. In Proceedings of the 40th International Symposium on Automation and Robotics in Construction (ISARC), pages 25–31, 2023

    work page 2023

  8. [8]

    Imple- mentation of an Augmented Reality AR Workflow for Human Robot Collaboration in Timber Prefabri- cation

    Ondrej Kyjanek, Bahar Al Bahar, Lauren Vasey, Benedikt Wannemacher, and Achim Menges. Imple- mentation of an Augmented Reality AR Workflow for Human Robot Collaboration in Timber Prefabri- cation. InProceedings of the 36th International Sym- posium on Automation and Robotics in Construction (ISARC), pages 1223–1230, 2019

    work page 2019

  9. [9]

    PhD thesis, ETH Zurich, 2020

    Arash Adel.Computational Design for Coopera- tive Robotic Assembly of Nonstandard Timber Frame Buildings. PhD thesis, ETH Zurich, 2020

    work page 2020

  10. [10]

    Challenges and potential for hu- man–robot collaboration in timber prefabrication

    Xiliu Yang, Felix Amtsberg, Michael Sedlmair, and Achim Menges. Challenges and potential for hu- man–robot collaboration in timber prefabrication. Automation in Construction, 160:105333, 4 2024. ISSN 09265805

    work page 2024

  11. [11]

    PhD thesis, ETH Zurich, 2015

    Tobias Bonwetsch.Robotically Assembled Brick- work: Manipulating Assembly Processes of Discrete Elements. PhD thesis, ETH Zurich, 2015

    work page 2015

  12. [12]

    Dragan, Kenton C.T

    Anca D. Dragan, Kenton C.T. Lee, and Siddhartha S. Srinivasa. Legibility and Predictability of Robot Mo- tion. In2013 8th ACM/IEEE International Confer- ence on Human-Robot Interaction (HRI), 2013

    work page 2013

  13. [13]

    Projected Aug- mented Reality Assembly Assistance System Sup- porting Multi-Modal Interaction.The International Journal of Advanced Manufacturing Technology, 123(3):1353–1367, 2022

    Jie Zhang, Shuxia Wang, Weiping He, Jianghong Li, Zhiwei Cao, and Bingzhao Wei. Projected Aug- mented Reality Assembly Assistance System Sup- porting Multi-Modal Interaction.The International Journal of Advanced Manufacturing Technology, 123(3):1353–1367, 2022

    work page 2022

  14. [14]

    Human-Robot Collaboration in Construction: Opportunities and Challenges

    Cynthia Brosque, Elena Galbally, Oussama Khatib, and Martin Fischer. Human-Robot Collaboration in Construction: Opportunities and Challenges. In 2020 International Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA), pages 1–8, 2020

    work page 2020

  15. [15]

    SangJun Ahn, SangUk Han, and Mohamed Al- Hussein. 2D Drawing Visualization Framework for Applying Projection-Based Augmented Reality in a Panelized Construction Manufacturing Facility: Proof of Concept.Journal of Computing in Civil Engineering, 33(5):04019032, 2019

    work page 2019

  16. [16]

    An Automated System for Projection of Inte- rior Construction Layouts.IEEE Transactions on Automation Science and Engineering, 16(4):1825– 1835, 2019

    Amir Degani, Wen Bo Li, Rafael Sacks, and Ling Ma. An Automated System for Projection of Inte- rior Construction Layouts.IEEE Transactions on Automation Science and Engineering, 16(4):1825– 1835, 2019

    work page 2019

  17. [17]

    Costa, Lu ´ıs Rocha, Pedro Malaca, Pedro Costa, Ant´onio P

    Pedro Tavares, Carlos M. Costa, Lu ´ıs Rocha, Pedro Malaca, Pedro Costa, Ant´onio P. Moreira, Armando Sousa, and Germano Veiga. Collaborative Welding System using BIM for Robotic Reprogramming and Spatial Augmented Reality.Automation in Construc- tion, 106:102825, 2019

    work page 2019

  18. [18]

    FBR Ltd. Hadrian. URLhttps://www.fbr.com. au/view/hadrian. access date: Mar 22, 2026

    work page 2026

  19. [19]

    Data- Driven Kinematic Control for Robotic Spatial Aug- mented Reality System with Loose Kinematic Spec- ifications.ETRI Journal, 38(2):337–346, 2016

    Ahyun Lee, Joo-Haeng Lee, and Jaehong Kim. Data- Driven Kinematic Control for Robotic Spatial Aug- mented Reality System with Loose Kinematic Spec- ifications.ETRI Journal, 38(2):337–346, 2016

    work page 2016

  20. [20]

    Tsai and Reimar K

    Roger Y. Tsai and Reimar K. Lenz. A New Technique for Fully Autonomous and Efficient 3D Robotics Hand/Eye Calibration. 5(3):345–358, 1989

    work page 1989

  21. [21]

    R. Tsai. A Versatile Camera Calibration Technique for High-Accuracy 3D Machine Vision Metrology Using Off-The-Shelf TV Cameras and Lenses.IEEE Journal on Robotics and Automation, 3(4):323–344, 1987

    work page 1987

  22. [22]

    Projection Defocus Anal- ysis for Scene Capture and Image Display

    Li Zhang and Shree Nayar. Projection Defocus Anal- ysis for Scene Capture and Image Display. InACM SIGGRAPH 2006, page 907, Boston, Massachusetts,

    work page 2006