Updating the art of joinery for the 21st century

Maria Larsson
Published in
5 min readOct 9, 2020


A 4-timber joint designed and fabricated using the Tsugite system. Left: Interface screenshots of closed (a) and open joint with tool path display (b). Right: Picture of 3-axis CNC-fabricated closed (c) and open joint (d).

Wooden joinery is a craft with a long history. Traditional joints exhibit an impressive variation of geometries. It is as if the joints have been topologically optimized for various applications without the use of a computer, by the passing down of expertise through generations of carpenters. And it is only the “genius carpenter” who possesses the knowledge of how to choose the right joint and how to craft it.

Three Computational Challenges

We identify three computational challenges to update the art of joinery for the 21st century.

  • To democratize the knowledge of how to design joints.
  • To use digital fabrication techniques for the manufacturing of joints.
  • To computationally replicate the historical evolution of joinery.

The Tsugite System

We address the above challenges by developing an interface, which will be presented at UIST 2020. We call the system “Tsugite” from the Japanese word for joinery.

Full-length introduction video to the Tsugite system (3 min)


Tsugite interface screenshot.

The interface is focused on the design of a single joint that is connecting two or more pieces of timbers. The design space is a 3D grid of voxels, and a joint can be edited by pushing and pulling on the faces. When we tested the system in a user study, even those without previous experience with 3D modeling were able to edit a joint.

Visual Feedback

There are a number of evaluation metrics that are being computed in the background each time the joint is edited. These are practical criteria for joinery, such as active sliding directions and durability in regard to the wood fiber direction. When there is a problem, the user receives feedback in form of colorful lines, fills, and textures that indicate the type and location of the problem. With the visual feedback, the user can avoid making mistakes, and perhaps also learn something about the geometrical criteria for joinery.

The eight evaluation criteria with visual feedback of the Tsugite system.

Suggestions & Gallery

For inexperienced users — or for expert users trying to design a particularly difficult joint—the interface offers further guidance in form of suggestions and a gallery. Suggestions show valid solutions within one edit distance from the current design. The gallery mode consists of pre-calculated valid joints for the current joint configuration.

Tsugite user guidance: Suggestions and gallery.

3-axis CNC Fabrication

3-axis CNC milling machine

The system inherently considers fabrication constraints and produces the tool path for manufacturing with a 3-axis CNC-machine. This method of fabrication is automated and very precise. Some people may say that it is a pity to replace a manual craft with automated fabrication—but there is also something satisfying about watching the machine perfectly cut out the joint you just designed.

Joint Samples

Four joints with the letters U-I-S-T.
Selection of joint samples designed and fabricated with the Tsugite system.

Furniture Prototype

A table with diagonal braces for stability. It is made from 10 wood bars connected by 14 joints.

So… could we solve the three computational challenges?

To a certain degree, yes. But there is still plenty to be done. Our joint geometries are limited to voxels, so we cannot make shapes such as a dovetail or a freeform curve. We also restrict ourselves to a 3-axis CNC-machine—a system made for 4 or more axis of motion could achieve other geometries. And we can generate feasible joints by a combinatorial search — but it is certainly possible both to increase the search space and to add even more criteria to select the best joints, not the least by more thorough structural analysis. Computers still have a long way to go to match the experience of a skilled carpenter. But it is our hope that the Tsugite system has contributed to somewhat reducing this gap.

Check out the project website to access a pre-print version of the paper, pre-recorded conference presentations, and more.

A Case for Sustainability

As a final note, I would like to emphasize the why. Why build with wood? And why use joints?

When it comes to our built environment, the materials we chose to build with have a large environmental impact. According to this UN report, the building and construction industry is responsible for almost 40% of worldwide carbon dioxide emissions. Wood is perhaps the only natural and renewable building material that we have. When searching for THE building material of the future, look no further than wood (presuming that the wood is responsibly resourced).

The technique of joinery adds further sustainability benefits to wood. When connecting timbers with joinery, as opposed to metal fixings, for example, material mixing is reduced. This is an advantage for sorting and recycling. Also, unglued joints can be taken apart without destroying the building components. This opens up potentials for buildings to be disassembled and reassembled somewhere else. Or for defect parts to be replaced. This flexibility of reuse and repair adds sustainability benefits to wood.

The current industry standard for wood construction is glue-laminated wood with metal fixtures. Maybe joinery updated for the 21st century can play a role in adding more sustainable and aesthetic values to the wood construction of the future?


Maria Larsson (1), Ph.D. Student
Hironori Yoshida (1,2), Researcher
Nobuyuki Umetani (1), Associate Professor
Takeo Igarashi (1), Professor

1: The University of Tokyo
2: Preferred Networks, Inc.


Maria Larsson, Hironori Yoshida, Nobuyuki Umetani, and Takeo Igarashi. 2020. Tsugite: Interactive Design and Fabrication of Wood Joints. In Proceedings of the 32nd Annual ACM Symposium on User Interface Software and Technology (UIST ‘20). Association for Computing Machinery, Virtual Event, USA. DOI:https://dx.doi.org/10.1145/3379337.3415899



Maria Larsson

Ph.D. Student at the User Interface Research Group, University of Tokyo. Wood enthusiast. Former architect designer.