MIT PRINTED A HINGE THAT FEELS ITS OWN BENDING - AND NO SENSOR GETS ADDED
A printed hinge that knows how far it bent, with no sensor added to it afterwards.
by Xiang Chang, Haiyang Yan, Stefanie Mueller and Jiaji Li
Open-hardwareRobotics
Built withSTM32
- difficulty
- ●●●●●
- time
- a serious weekend-plus
- license
- license not specified
- repo
- repo 0 stars
●●●●● · a serious weekend-plus · license not specified · 0 stars · repo
WHAT YOU’LL NEED
Jump to section
COMPAREE VERDICT
X-Hinges is one of those ideas that makes you stop and think: the hinge is the sensor, and it all comes out of the printer in one piece. The team (MIT CSAIL with Tianjin and Zhejiang universities) co-prints two conductive filaments of very different conductivity inside a soft TPU body, so when the hinge bends, twists or compresses, the sensing elements deform and their resistance changes; a precision readout board and data-driven models turn that into motion. Each of the three degrees of freedom gets its own sensing element. The paper demonstrates a teleoperation glove driving a robotic hand, an orca-shaped game controller, a Kresling origami lamp, and a tactile matrix that recognises fruit. The hard truth: this is a research publication accepted to UIST 2026, not a ready-to-build project. There are no files, no repository, no firmware and no tool download. The paper gives you the structure, the materials, the readout architecture and the design approach, but you would be rebuilding the geometry, the electronics and the training pipeline yourself. You also need a multi-material FDM printer that can handle two conductive filaments plus TPU in a single job. If you have that setup and the patience to work from a paper, this is a genuinely novel sensing method. If you are hoping for a quick weekend print, this is the wrong project.
IN THE REPO
GOOD TO KNOW
- —This is a research paper accepted to UIST 2026, not a released product. The paper (arXiv 2609.11077) describes the method in full detail.
- —No repository, STLs, Gerber files or firmware have been published.
- —The paper includes the design tool architecture and sensor model equations, but the tool itself is not released.
- —Readout hardware is custom: a voltage follower, transimpedance amplifier and ADS1256 ADC, 4 channels per board (scalable to 32) at 50 Hz. The paper shows the circuit composition and a board photo, but no board files are released.
- —Requires a consumer multi-material FDM printer (the paper names the Prusa XL, Bambu H2D and Snapmaker U1) co-printing two conductive filaments with non-conductive TPU.
- —The paper is published under CC BY-NC-ND 4.0; no hardware licence exists because no files are released. Contact the authors before replicating commercially.
Parts to buy
5 itemsFrom our check of the build. Exact quantities and part numbers are in the creator’s BOM.
Can I build this?
Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.
More builds like this
All projectsGallery
Start here
Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.
- 1.Read the paper (The only resource. It describes the sensor model, design approach, and readout hardware in full academic detail.)
- 2.Check if the tool or files are released(No repository or tool download has been published yet. Contact the authors if you want to replicate this.)
- 3.Confirm your printer can handle it(You need a multi-material FDM machine for two conductive filaments plus non-conductive TPU — the authors name the Prusa XL, Bambu H2D and Snapmaker U1.)
KNOWN ISSUES
- The paper is not a released project. No files, no repository, no firmware, no tool download. You are reverse-engineering from figures and equations.
- Requires a multi-material FDM printer that can handle three filaments in one job — the authors name the Prusa XL, Bambu H2D and Snapmaker U1. A single-material printer will not do it.
- The readout board is custom. The paper includes its complete schematics (Appendix A.5, Figure 22) and a block diagram, but no board layout or Gerber files — you will lay out and build the board yourself.
- Two conductive filaments with very different resistivities are required — the paper uses ABC3D ESD Grade TPU (about 1.23 × 10⁴ Ω·cm) for sensing and a roughly 8.75 Ω·cm filament for traces. Substitutes will change the signal.
- Calibration is non-trivial. The authors turn raw resistance into motion with data-driven regression models trained on many recorded samples, so you would need to collect your own training data for your prints and build that pipeline yourself.
- The paper is CC BY-NC-ND 4.0 and no design files carry their own licence — if you plan to commercialise or publicly share derivatives, contact the authors first.
Can I print this on my Prusa or Bambu?
Yes, if it is a multi-material model. The authors say the three-material print (two conductive filaments plus non-conductive TPU) works on any consumer multi-material FDM printer and name the Prusa XL, Bambu H2D and Snapmaker U1. Single-nozzle machines without a tool changer or multi-material system are out.
Where are the STL files?
There are none yet. This is a research paper accepted to UIST 2026, not a released project. The paper describes the method in full, but you are reconstructing the geometry from figures and text.
What conductive filament should I buy?
The paper uses ABC3D ESD Grade conductive TPU (about 1.23 × 10⁴ Ω·cm) for the sensing elements and a much more conductive filament (about 8.75 Ω·cm) for the traces, inside a non-conductive TPU body. Match those resistivities as closely as you can.
Is the design tool available?
No. The paper describes the tool's architecture and how it generates sensing layouts, but the tool itself is not released. You would be implementing it from the description.
Can I use this commercially?
Not without asking. The paper is published under CC BY-NC-ND 4.0 and no design files or separate hardware licence have been released, so contact the authors before commercialising anything based on it.
Community builds
No community builds yet — be the first, we feature the best ones.
Discussion1
FROM THE COMPAREE TEAM
The paper shows a printed orca as a game controller and a tactile pad that knows when you put an apple on it — both from resistance changes in deforming traces, no discrete sensors. If you had access to the design tool and a multi-material printer, what would you embed sensing into first?
Xiang Chang, Haiyang Yan, Stefanie Mueller and Jiaji Li
The team spans MIT CSAIL's HCI Engineering group, Tianjin University and Zhejiang University. The work was accepted to UIST 2026 (ACM Symposium on User Interface Software and Technology), one of the top venues for novel interaction techniques. The research focuses on embedding sensing directly into 3D-printed structures, eliminating the need for post-assembly sensor attachment.
DISCLAIMER
- Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
- Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
- Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
CompareeTEAM24d agoedited
Practical notes from our verification: this is a research paper (arXiv 2609.11077, accepted at ACM UIST 2026) from researchers at MIT CSAIL, Tianjin University and Zhejiang University, not a released project — as of our check there is no repository, STL files, firmware or download for the design tool, so you would be working from the paper's figures, equations and description of the readout hardware. The paper is published under CC BY-NC-ND 4.0, with no separate licence for the design itself, so contact the authors before replicating it for anything commercial. The key requirement is a multi-material FDM printer that can co-print two conductive filaments of different conductivities in one job; the paper names consumer machines such as the Prusa XL, Bambu H2D and Snapmaker U1, and uses ABC3D ESD-grade conductive TPU for the sensing elements. The demonstrations are genuinely impressive, but this is a research prototype, not a weekend project. Correction (4 October 2026): we re-checked this page line by line against the project's own repository, documentation and videos, and fixed errors in earlier versions.