A Machine Learning Approach to Contact Localization in Variable Density Three-Dimensional Tactile Artificial Skin
Abstract
In artificial tactile sensing, accurately localizing contact points on artificial skin is an important function. The performance of existing contact localization methods is constrained by the specific geometry and sensor locations used in the artificial skin, which limits their ability to be used on 3D surfaces. This paper studies the contact localization on an artificial skin embedded with mutual capacitance tactile sensors, arranged non-uniformly in a semi-conical 3D geometry. A fully-connected neural network is trained to localize the touching points on the embedded tactile sensors. The precision exhibits a standard deviation of localization error of 6 ± 3 mm. This research contributes a versatile tool and robust solution for contact localization in artificial tactile systems.
Cite
Text
Murray et al. "A Machine Learning Approach to Contact Localization in Variable Density Three-Dimensional Tactile Artificial Skin." NeurIPS 2024 Workshops: WTP, 2024.Markdown
[Murray et al. "A Machine Learning Approach to Contact Localization in Variable Density Three-Dimensional Tactile Artificial Skin." NeurIPS 2024 Workshops: WTP, 2024.](https://mlanthology.org/neuripsw/2024/murray2024neuripsw-machine/)BibTeX
@inproceedings{murray2024neuripsw-machine,
title = {{A Machine Learning Approach to Contact Localization in Variable Density Three-Dimensional Tactile Artificial Skin}},
author = {Murray, Mitchell and Zhang, Yutong and Kohlbrenner, Carson and Escobedo, Caleb and Dunnington, Thomas and Stevenson, Nolan and Correll, Nikolaus and Roncone, Alessandro},
booktitle = {NeurIPS 2024 Workshops: WTP},
year = {2024},
url = {https://mlanthology.org/neuripsw/2024/murray2024neuripsw-machine/}
}