GRAFT


a relational data framework for robotic tim-ber fabrication and assembly
Abstract. This paper presents a graph-based data structure for robotic timber fabrication that unifies reciprocal structural information, assembly sequencing, and inverted robotic path planning within a shared relational framework. Current fabrication workflows in the Architecture, Engineering, and Construction (AEC) industry are dominated by linear and opaque data exchanges, limited interoperability within current industry, and limited adaptability to new designs. By contrast, GRAFT (Graph-Based Robotic Assembly and Fabrication for Timber) formalizes fabrication dependencies as a graph network, linking data nodes and edges to material stock and generating fabrication data through explicit edge relationships. This structure allows design intent, material properties, and fabrication logic to co-evolve dynamically throughout robotic production. A prototype implementation demonstrates how fabrication data derived from the graph can be decomposed into modular machining operations and executed through a multi-tool robotic setup. The approach integrates standard industrial data protocols while maintaining compatibility with emerging BIM-graph data frameworks. Results from a reciprocal timber frame case study highlight the framework’s capacity to improve data transparency, enable adaptive assembly sequencing, and optimize robotic timber fabrication. The proposed method contributes to developing scalable, machine-driven data infrastructures that connect design, material, and fabrication intelligence promoting adaptive, automation-ready timber production. Theme D: Robots + Emerging Methodologies.

Keywords: Robotic fabrication, Graph-based data models, Digital timber construction, Relational data structures, BIM integration.