34th Annual Conference of the International Group for Lean Construction (IGLC 34) - Singapore, Singapore - 2026

https://doi.org/10.24928/2026/0317

LLM-based action learning model for construction robotics: an industry-academia co-design

Atefeh Aali1, Ruoyu Yan2, Reza Maalek3 & Xinyu Zheng4

1Research Associate/ MSc./ Department of Civil Engineering/ Karlsruhe Institute of Technology (KIT)/ Institute of Technology and Management in Construction, Karlsruhe, Germany, [email protected], orcid.org/0009-0005-7058-9265
2Master’s student/ BSc./ Department of Civil Engineering/ Karlsruhe Institute of Technology (KIT)/ Institute of Technology and Management in Construction, Karlsruhe, Germany, [email protected], orcid.org/0009-0009-0784-1850
3Senior Lecturer in Digital Engineering in Infrastructure and Built Environment/ Faculty of Engineering and Science/ University of Greenwich, London, United Kingdom, [email protected], orcid.org/0000-0001-6825-2691
4Master’s student/ BSc./ Department of Civil Engineering/ Karlsruhe Institute of Technology (KIT)/ Institute of Technology and Management in Construction, Karlsruhe, Germany, [email protected], orcid.org/0009-0009-7962-4839

Abstract

The construction industry continues to face challenges in adopting advanced technologies, such as construction robotics, from research to practice. From a lean management perspective, this constitutes wasted scientific effort and limited gain in efficiency and productivity of construction processes. Traditionally, interactive workshops that bring together academic and industry practitioners are commonly used to foster knowledge exchange; however, they often lack structured mechanisms to transform diverse expert insights into actionable solutions to address real-world industry challenges. This paper addresses these limitations by introducing a novel AI-assisted workshop method to prioritize critical problems in construction robotics and develop actionable solution pathways. The approach was implemented during a two-day workshop in Germany, which hosted over 100 leaders in construction robotics from academia and industry. An LLM-supported workflow was used to synthesize expert inputs and the data from the literature to identify key areas with potential for further research. These problems were then prioritized through a democratic live polling process involving all participants. Experts were subsequently divided into breakout groups based on predefined criteria and developed solution approaches using their domain knowledge, which were presented in a final plenary session. This paper reports key lessons learned, including outcomes from expert presentations, throughout this process.

Keywords

Lean construction, collaboration, action learning, large language models (LLMs), construction robotics.

Files


Full Text
 

Reference

Download: BibTeX | RIS Format

Reference in APA 7th edition format:

Aali, A., Yan, R., Maalek, R. & Zheng, X.. (2026). LLM-based action learning model for construction robotics: an industry-academia co-design. In Hamzeh, F., Poshdar, M., & Garcia-Lopez,, N. P. (Eds.), Proceedings of the 34th Annual Conference of the International Group for Lean Construction (IGLC 34) (pp. 1358–1369). https://doi.org/10.24928/2026/0317

Shortened reference for use in IGLC papers:

Aali, A., Yan, R., Maalek, R. & Zheng, X.. (2026). LLM-based action learning model for construction robotics: an industry-academia co-design. IGLC34. https://doi.org/10.24928/2026/0317

 

Founded in 1993, the IGLC is an international network of researchers from practice and academia in architecture, engineering, and construction (AEC) who feel that the practice, education, and research of the AEC industry have to be radically renewed in order to respond to the global challenges ahead.

2015 © All Rights Reserved.