What is it about?

If I download a laser-cut model, I would like to be able to make changes to it so that it becomes useful for me. If I have a VR headset model, as shown in Figure 1, I may want to change the distance between the lenses and the image to fit it to my prescription because I have glasses. To make such a chance is normally hard, because the model is shared as a 2D cutting plan. I need to find out how the different plates are connected and imagine the change I want to make in 3D to then modify the 2D plates. In a 3D modeling environment it is actually surprisingly easy: I just stretch the model... We built assembler3, a software tool that makes it easy to turn a 2D cutting plan into a 3D model. It uses a 5-step algorithm to analyse the 2D cutting plan and then empowers users to quickly puzzle together the 3D model in kyub, a 3D editor for laser cutting. Users can immediately modify the resulting model. We ran a study to compare this workflow to the manual workflow in 2D. In the manual condition, it took on average 24 minutes to modify the VR headset and 11/13 models failed to fabricate when we sent them to the laser cutter. With assembler^3, this workflow was 10x faster and all models continued to work as a result

Featured Image

Why is it important?

At the moment it is really hard to productively share models for laser cutting. The 2D cutting plans only work well on the exact same laser cutter and for the exact same person. If I reproduce the VR headset that was modeled by somebody else, it will likely fail because my laser removes more/less material than the original. Even harder is making any change to the model because of the tedious workflow described before. I call these "portability" problems, they are problems that arise when trying to make the model work across different use cases. These portability problems arise because we try to make something in 3D but we represent it in 2D cutting plans (because that's what the laser cuts). I propose we should use 3D all the way. This is why my lab developed kyub (www.kyub.com): a 3D modeling environment for laser cutting. But there are so many models out there already that exist in the form of 2D cutting plans and experts spent entire careers learning how to best model those. Assembler^3 allows them to upgrade those models to 3D models, which are now more valuable and useful for a broader audience.

Perspectives

After my work on SpringFit (https://dl.acm.org/doi/10.1145/3332165.3347930) and KerfCanceler (https://dl.acm.org/doi/10.1145/3379337.3415895) in which I focused on making the 2D cutting plans generic by modifying the joints and mechanisms so that they work on any laser cutter, I decided to try a radically different approach. The cutting plans of SpringFit and KerfCanceler are great because everyone can fabricate them right away, but this portability comes at a cost: additional incisions in the models. With Assembler^3 I found a way to tackle the problem in the model itself by going from 2D to 3D. This level of abstraction is crucial in handling different fabrication machines. On a higher level I am solving a hardware problem (kerf) in software (by modifying the model) so that everyone on the planet can benefit from the solution. The answer to the problem is similar to how compilers allow you to program in a high level language which is eventually compiled to assembler code that works on that particular machine.

Thijs Roumen
Hasso Plattner Institute

Read the Original

This page is a summary of: Assembler3: 3D Reconstruction of Laser-Cut Models, May 2021, ACM (Association for Computing Machinery),
DOI: 10.1145/3411764.3445453.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page