Democratizing Construction Robotics: A Neuro-Symbolic Framework for Semantic Spatial Adaptation and Intuitive Control
Journal
Proceedings of the International Symposium on Automation and Robotics in Construction
Start Page
994
End Page
1001
ISSN
2413-5844
ISBN (of the container)
978-064583223-5
ISBN
[9780645832235]
Date Issued
2026-06-22
Author(s)
Abstract
This paper presents a neuro-symbolic framework for intuitive robot control in construction, addressing the technical complexity barrier that limits robotics adoption on construction sites. The framework comprises two components: a Neural Component (large language model) that interprets natural language instructions and generates high-level spatial primitives, and a Symbolic Component that transforms these primitives into precise robot coordinates through deterministic computation. This separation ensures that the language model cannot produce invalid robot motions, as all outputs are validated through geometric calculation. The framework was evaluated through proof-of-concept experiments involving brick wall stacking (9 objects, 54 motions) and timber beam arrangement (5 objects, 30 motions) in a physics-based simulation. Results demonstrated over 90% reduction in task specification time compared to manual coordinate entry. A human-in-the-loop verification process enables users to validate planned motions through 3D visualization and refine tasks through iterative natural language dialogue, typically converging within 3-5 conversation rounds. The framework lowers the barrier to robot programming by enabling construction practitioners to express spatial intent using domain-familiar vocabulary rather than coordinates or code.
Event(s)
43rd International Symposium on Automation and Robotics in Construction, ISARC 2026,22 June 2026 - 26 June 2026,Singapore
Subjects
Construction Automation
Human-Robot Collaboration
Large Language Models (LLM)
Natural Language Interface
Publisher
International Association for Automation and Robotics in Construction (IAARC)
Type
conference paper