Skip to content
RobothecaBeta
Log in

Sparsh and Sparsh-X (touch)

  • AI model
  • Perception

Sparsh-X and Sparsh-Skin are encoder backbones that learn touch representations, fusing Digit 360 tactile modalities or embedding magnetic skin data for dexterous hands.

  • Non-commercial
  • Non-commercial weights
  • Stale

Made by Meta, United States (checked on github.com/facebookresearch (opens in a new tab)).

Source on GitHub: github.com/facebookresearch/sparsh-multisensory-touch (opens in a new tab). Facts read at commit 219ee2f (opens in a new tab).

Licenses

Code
CC-BY-NC-4.0 (opens in a new tab)

Read by Robotheca from LICENSE.md (opens in a new tab) (CC-BY-NC-4.0) at commit 219ee2f (opens in a new tab), on 16 Oct 2026.

Commercial use
No

The code and weights licenses do not allow commercial use.

Labels

  • Non-commercialThe code license (CC-BY-NC-4.0) does not allow commercial use.
  • Non-commercial weightsThe weights license (CC-BY-NC-4.0) does not allow commercial use.
  • StaleNo commit in 12 months: the newest is from 20 Aug 2025.

What it does

  • Sparsh-X fuses tactile image, audio, IMU and pressure signals from the Digit 360 sensor into one representation.
  • Sparsh-Skin provides embeddings for tactile data from magnetic skins covering dexterous hands.
  • Pretrained checkpoints for Sparsh-X (all modalities and image-only) and Sparsh-Skin are hosted on Hugging Face.
  • Includes a forward-pass example for loading a pretrained encoder and computing tactile embeddings on dummy input.
  • Training and downstream task configs use Hydra, with self-supervised DINO-style pretraining and task-specific evaluation.
  • Provides pretraining datasets for both sensors, with scripts to convert rosbag recordings and format them for training.

Written by Robotheca from the project's README (opens in a new tab) at commit 219ee2f.

Getting started

Difficulty
Advanced

Maintenance

Status
Stale: no commit in 12 months
Newest commit
20 Aug 2025 on main
Latest release
None published on GitHub

Checked on GitHub on 16 Oct 2026.

Links

Robots on Robotheca that use it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.