Âé¶¹´«Ã½

Research & Art

When Tree Types Matter: AI-Assisted Joint Design for Diverse Wood Species

This project focuses on often overlooked tree types and sheds light on how they can inform furniture design using AI-powered optimization.
Main image for art-ai-fact project by Yoshida and Larson
A stack of timber, presenting the unique characteristics of individual planks of wood. (picture: Maria Larsson)

Project authors: (Future University Hakodate) and (The University of Tokyo). Both will visit Aalto University as visiting researchers during the project.

This project focuses on often overlooked tree types and sheds light on how they can inform furniture design using AI-powered optimization. Traditionally, furniture designs tend toward uniformity, with identical products often masking the unique qualities of various wood species. However, the recent global wood shortage has compelled manufacturers to switch from imported timber to various locally sourced woods—each with distinct mechanical properties that challenge fixed industrial production lines. Taking this as a design opportunity, this project aims to develop an AI-assisted design workflow that updates and optimizes joint parts in furniture according to material uniqueness. The role of AI in this project is twofold: image-based material analysis, and numerical evaluation of aesthetic quality.

Recent developments:

UpJoint
As a first outcome of this project, we developed UpJoint, a computational design system that adapts a base chair design to wood species with distinct mechanical properties. The system uses a hybrid construction approach combining wooden structural elements with 3D-printed joints, enabling flexible adaptation across materials without changes to existing manufacturing setups. UpJoint defines a discrete, structured design space of adaptive strategies, including reinforcement and the insertion of additional structural members, allowing a wide range of feasible design variations under consistent fabrication constraints. In addition to geometry- and material-based estimates of structural strength and production cost, the system incorporates learned aesthetic preferences derived from human pairwise comparison data, which are aggregated into a global ranking model. This work was published as a full paper in the proceedings of the ACM Symposium on Computational Fabrication (SCF) 2025.
 

Ren Sato, Maria Larsson, and Hironori Yoshida. 2025. UpJoint: Updating 3D-Printed Joints for Various Wood Species. In Proceedings of the ACM Symposium on Computational Fabrication (SCF '25). Association for Computing Machinery, New York, NY, USA, Article 12, 1–13.

art-ai-fact

A developing collection of design projects with an AI element to them, building understanding of AI's contribution to design at Aalto ARTS.

Read more
Evolumination at DDW 2023, by Kollegi.
  • Updated:
  • Published:
Share
URL copied!