Robotically Produced Timber Dowel Double-Curvature Discrete Shell
Year: 2025
Type: conference-paper
Abstract: This paper presents the development and realization of a double‑curvature timber discrete shell built with a dry‑assembled, dowel‑joined system and an integrated pipeline that combines computational design, parametric structural analysis, multi‑axis robotic milling, and augmented‑reality‑guided production and assembly. An iterative macro‑to‑micro workflow enables form‑finding, aligning architectural geometry with structural performance, fabrication precision, and assembly logic. Adaptive tessellation—implemented through a recursive method that respects the constraints and potentials of timber‑dowel connections—divides the surface into a mesh of non‑convex octagons, from which discrete timber components of varied length, thickness, and joint orientation are derived. Iterative parametric structural simulations across multiple joint‑and‑support degree‑of‑freedom configurations steer material allocation and inform timber and dowel distribution. A 7‑axis robotic milling setup delivers accuracy fabrication of these non‑repetitive geometries while avoiding kinematic singularities; on‑site assembly is streamlined by spatially aligned holographic guidance, eliminating reliance on conventional drawings, and a tertiary locking pin prevents dowel rotation to secure each joint. The process culminates in a full‑scale pavilion comprising 181 unique timber members and 330 hardwood dowels, demonstrating a data‑ and resource‑driven design‑to‑production workflow for scalable timber construction. Following these results, the paper concludes by outlining future work on multi‑actor human–robot collaborative assembly processes for timber construction, with an emphasis on design for assembly and disassembly at larger scales.
Authors: Bahar Bagheri, Sina Mostafavi, Edgar Humberto Montejano Hernandez, Mark Andrew Segovia, Caleb Randall Scott, Uijin Lee
Institution: Texas Tech University
Language: English
Subfield: Architecture
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Funder: National Science Foundation
OpenAlex ID: W4416200180