Abstract

Skin-like tactile sensors provide robots with rich feedback related to the force distribution applied to their soft surface. The complexity of interpreting raw tactile information has driven the use of machine learning algorithms to convert the sensory feedback to the quantities of interest. However, the lack of ground truth sources for the entire contact force distribution has mainly limited these techniques to the sole estimation of the total contact force and the contact center on the sensor's surface. The method presented in this article uses a finite element model to obtain ground truth data for the three-dimensional force distribution. The model is obtained with state-of-the-art material characterization methods and is evaluated in an indentation setup, where it shows high agreement with the measurements retrieved from a commercial force-torque sensor. The proposed technique is applied to a vision-based tactile sensor, which aims to reconstruct the contact force distribution purely from images. Thousands of images are matched to ground truth data and are used to train a neural network architecture, which is suitable for real-time predictions.

Highlights

  • A growing number of applications require robots to interact with the environment [1] and with humans [2]

  • These maps are generally retrieved by means of supervised learning techniques, which fit a model to a large amount of labeled data, i.e., sensory data paired with the corresponding ground truth

  • The strategy followed in this article exploits finite element method (FEM) simulations to obtain ground truth data for the contact force distribution applied to the soft surface of a tactile sensor

Read more

Summary

INTRODUCTION

A growing number of applications require robots to interact with the environment [1] and with humans [2]. The sensor used here for the evaluation of the proposed approach is based on an RGB camera (which retains a small size) that tracks a dense spread of particles within a soft gel, see for example Fig. 1 This design is presented in [9] and shows performance advantages over sparse marker tracking, and ease of manufacture, without any assumptions about the surface shape. The strategy followed in this article exploits FEM simulations to obtain ground truth data for the contact force distribution applied to the soft surface of a tactile sensor. Soft elastomers are often modeled as hyperelastic materials [26], and finding a suitable model formulation and corresponding parameters generally necessitates experimental data from both uniaxial and biaxial stress states [27], [28] To this end, a large-strain multiaxial characterization of the two most compliant materials, the Ecoflex GEL and the Elastosil 25:1, is performed. To assess the influence of aging, additional UA test pieces of the same sheets were kept at room temperature and tested several weeks after fabrication

EXPERIMENTAL DATA ANALYSIS
GENERATING A DATASET
Findings
CONCLUSION
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call