Self-supervised perception for tactile skin covered dexterous hands

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Sharma, Akash, Higuera, Carolina, Bodduluri, Chaithanya Krishna, Liu, Zixi, Fan, Taosha, Hellebrekers, Tess, Lambeta, Mike, Boots, Byron, Kaess, Michael, Wu, Tingfan, Hogan, Francois Robert, Mukadam, Mustafa
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910948002889728
author Sharma, Akash
Higuera, Carolina
Bodduluri, Chaithanya Krishna
Liu, Zixi
Fan, Taosha
Hellebrekers, Tess
Lambeta, Mike
Boots, Byron
Kaess, Michael
Wu, Tingfan
Hogan, Francois Robert
Mukadam, Mustafa
author_facet Sharma, Akash
Higuera, Carolina
Bodduluri, Chaithanya Krishna
Liu, Zixi
Fan, Taosha
Hellebrekers, Tess
Lambeta, Mike
Boots, Byron
Kaess, Michael
Wu, Tingfan
Hogan, Francois Robert
Mukadam, Mustafa
contents We present Sparsh-skin, a pre-trained encoder for magnetic skin sensors distributed across the fingertips, phalanges, and palm of a dexterous robot hand. Magnetic tactile skins offer a flexible form factor for hand-wide coverage with fast response times, in contrast to vision-based tactile sensors that are restricted to the fingertips and limited by bandwidth. Full hand tactile perception is crucial for robot dexterity. However, a lack of general-purpose models, challenges with interpreting magnetic flux and calibration have limited the adoption of these sensors. Sparsh-skin, given a history of kinematic and tactile sensing across a hand, outputs a latent tactile embedding that can be used in any downstream task. The encoder is self-supervised via self-distillation on a variety of unlabeled hand-object interactions using an Allegro hand sensorized with Xela uSkin. In experiments across several benchmark tasks, from state estimation to policy learning, we find that pretrained Sparsh-skin representations are both sample efficient in learning downstream tasks and improve task performance by over 41% compared to prior work and over 56% compared to end-to-end learning.
format Preprint
id arxiv_https___arxiv_org_abs_2505_11420
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Self-supervised perception for tactile skin covered dexterous hands
Sharma, Akash
Higuera, Carolina
Bodduluri, Chaithanya Krishna
Liu, Zixi
Fan, Taosha
Hellebrekers, Tess
Lambeta, Mike
Boots, Byron
Kaess, Michael
Wu, Tingfan
Hogan, Francois Robert
Mukadam, Mustafa
Robotics
We present Sparsh-skin, a pre-trained encoder for magnetic skin sensors distributed across the fingertips, phalanges, and palm of a dexterous robot hand. Magnetic tactile skins offer a flexible form factor for hand-wide coverage with fast response times, in contrast to vision-based tactile sensors that are restricted to the fingertips and limited by bandwidth. Full hand tactile perception is crucial for robot dexterity. However, a lack of general-purpose models, challenges with interpreting magnetic flux and calibration have limited the adoption of these sensors. Sparsh-skin, given a history of kinematic and tactile sensing across a hand, outputs a latent tactile embedding that can be used in any downstream task. The encoder is self-supervised via self-distillation on a variety of unlabeled hand-object interactions using an Allegro hand sensorized with Xela uSkin. In experiments across several benchmark tasks, from state estimation to policy learning, we find that pretrained Sparsh-skin representations are both sample efficient in learning downstream tasks and improve task performance by over 41% compared to prior work and over 56% compared to end-to-end learning.
title Self-supervised perception for tactile skin covered dexterous hands
topic Robotics
url https://arxiv.org/abs/2505.11420