# Meta Reality Labs details Tactus, a textile glove for capturing hand manipulation data

> Meta Reality Labs Research describes Tactus, a textile glove with tactile and pose sensing for recording fine-grained hand manipulation data.

- Status: official (From the company, authors or authority themselves, with a link to their own page.)
- Topic: Research
- Published: 2026-11-09T13:31:34+00:00
- Updated: 2026-11-09T14:12:32.767821+00:00
- Web page: https://ai-news.ikangai.com/story/meta-reality-labs-details-tactus-a-textile-glove-for-capturing-hand-manipulation/
- Seen in: Import AI (Jack Clark)
- Change: research (Research)
- Applies to: Tactus

## What changed

Meta Reality Labs published an arXiv paper describing Tactus, a textile glove that combines capacitive tactile sensors with IMU pose tracking to record hand manipulation data.

Meta Reality Labs Research has published details of Tactus, a textile glove that combines capacitive tactile sensors with inertial-measurement-unit pose tracking to record how human hands handle objects. The paper targets a gap in robot learning. Cameras can see a hand, but they often lose detail once the hand blocks the object it holds, and that detail is what training data needs.

Meta lists three pillars in its arXiv paper. The first is a thin, conformal 3D-knit form factor with a 1.5 mm fingertip stackup, meant to avoid changing how the wearer grips things. The second is a set of capacitive tactile sensors placed at regions chosen from what the authors call empirical contact studies. The third is IMU-fused pose tracking, which the authors say maintains accuracy through hand-object occlusions, where manipulation data is most valuable. Together, these are meant to provide calibrated fingertip force sensing and hand tracking that keeps working under occlusion.

The abstract gives the numbers behind those claims. Tactus resolves a minimum detectable force of 0.59 N at 95% confidence, which the authors say is 1.6 to 2.7 times lower than two commercially available tactile gloves they tested against. It reports 0.87 cross-participant consistency. Its pose tracking sustains 11.0 mm fingertip keypoint relative error during manipulation, when occlusion is heaviest. The authors also say they introduced standardized benchmarking protocols for tactile force regression and hand pose estimation, which allows the cross-glove comparison. Those comparisons come from the authors' own evaluation, so they are the authors' results rather than independent measurements.

The paper says the combined signals support object property estimation, slip detection and recording of hand-object interaction. It also includes a per-session calibration step for correcting subject-dependent differences. The abstract does not say how many participants were involved in the cross-participant consistency measurement. It describes the glove's validation in terms of data fidelity rather than any trained robot policy that used the data.

## Why it matters

Better manipulation data can help train robots and dexterous AI systems, so teams building in robotics should track this data-collection approach.

## Original source

- [(2610.39076) Tactus: Tactile & Pose Sensing Glove for Natural Dexterity Data Collection](https://arxiv.org/abs/2610.39076)

## Further reporting

- [Import AI 480: Self-sustaining space economies; AI for science progress, and Meta builds perfect hands for a superintelligence](https://importai.substack.com/p/import-ai-480-self-sustaining-space)

---

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it.

Cite as: IKANGAI AI News, "Meta Reality Labs details Tactus, a textile glove for capturing hand manipulation data", 2026-11-09, https://ai-news.ikangai.com/story/meta-reality-labs-details-tactus-a-textile-glove-for-capturing-hand-manipulation/
