Investigating Active Sampling for Hardness Classification with Vision-Based Tactile Sensors

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Junyi, Kshirsagar, Alap, Heller, Frederik, Andreu, Mario Gómez, Belousov, Boris, Schneider, Tim, Lin, Lisa P. Y., Doerschner, Katja, Drewing, Knut, Peters, Jan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910953668345856
author Chen, Junyi
Kshirsagar, Alap
Heller, Frederik
Andreu, Mario Gómez
Belousov, Boris
Schneider, Tim
Lin, Lisa P. Y.
Doerschner, Katja
Drewing, Knut
Peters, Jan
author_facet Chen, Junyi
Kshirsagar, Alap
Heller, Frederik
Andreu, Mario Gómez
Belousov, Boris
Schneider, Tim
Lin, Lisa P. Y.
Doerschner, Katja
Drewing, Knut
Peters, Jan
contents One of the most important object properties that humans and robots perceive through touch is hardness. This paper investigates information-theoretic active sampling strategies for sample-efficient hardness classification with vision-based tactile sensors. We evaluate three probabilistic classifier models and two model-uncertainty-based sampling strategies on a robotic setup as well as on a previously published dataset of samples collected by human testers. Our findings indicate that the active sampling approaches, driven by uncertainty metrics, surpass a random sampling baseline in terms of accuracy and stability. Additionally, while in our human study, the participants achieve an average accuracy of 48.00%, our best approach achieves an average accuracy of 88.78% on the same set of objects, demonstrating the effectiveness of vision-based tactile sensors for object hardness classification.
format Preprint
id arxiv_https___arxiv_org_abs_2505_13231
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Investigating Active Sampling for Hardness Classification with Vision-Based Tactile Sensors
Chen, Junyi
Kshirsagar, Alap
Heller, Frederik
Andreu, Mario Gómez
Belousov, Boris
Schneider, Tim
Lin, Lisa P. Y.
Doerschner, Katja
Drewing, Knut
Peters, Jan
Robotics
Machine Learning
One of the most important object properties that humans and robots perceive through touch is hardness. This paper investigates information-theoretic active sampling strategies for sample-efficient hardness classification with vision-based tactile sensors. We evaluate three probabilistic classifier models and two model-uncertainty-based sampling strategies on a robotic setup as well as on a previously published dataset of samples collected by human testers. Our findings indicate that the active sampling approaches, driven by uncertainty metrics, surpass a random sampling baseline in terms of accuracy and stability. Additionally, while in our human study, the participants achieve an average accuracy of 48.00%, our best approach achieves an average accuracy of 88.78% on the same set of objects, demonstrating the effectiveness of vision-based tactile sensors for object hardness classification.
title Investigating Active Sampling for Hardness Classification with Vision-Based Tactile Sensors
topic Robotics
Machine Learning
url https://arxiv.org/abs/2505.13231