Unsupervised preprocessing for Tactile Data

TL;DR

This paper introduces an unsupervised learning method to transform complex tactile data into compact representations, facilitating control and calibration without requiring ground truth data, thereby enhancing tactile sensor usability.

Contribution

It presents a novel unsupervised preprocessing approach that creates useful tactile data representations for control and calibration, reducing reliance on labeled data.

Findings

Latent representations predict key tactile features accurately.

The method enables effective tactile sensor calibration with minimal data.

Compact representations improve tactile-based control tasks.

Abstract

Tactile information is important for gripping, stable grasp, and in-hand manipulation, yet the complexity of tactile data prevents widespread use of such sensors. We make use of an unsupervised learning algorithm that transforms the complex tactile data into a compact, latent representation without the need to record ground truth reference data. These compact representations can either be used directly in a reinforcement learning based controller or can be used to calibrate the tactile sensor to physical quantities with only a few datapoints. We show the quality of our latent representation by predicting important features and with a simple control task.