Learning Audio Feedback for Estimating Amount and Flow of Granular Material - Robotics Institute Carnegie Mellon University

Learning Audio Feedback for Estimating Amount and Flow of Granular Material

Samuel Clarke, Travers Rhodes, Christopher G. Atkeson, and Oliver Kroemer
Conference Paper, Proceedings of Conference on Robot Learning, Vol. 87, pp. 529-550, October, 2018

Abstract

Granular materials produce audio-frequency mechanical vibrations in air and structures when manipulated. These vibrations correlate with both the nature of the events and the intrinsic properties of the materials producing them. We therefore propose learning to use audio-frequency vibrations from contact events to estimate the amount and flow of granular materials during scooping and pouring tasks. We evaluated multiple deep and shallow learning frameworks on a dataset of 13,750 shaking and pouring samples across five different granular materials. Our results indicate that audio is an informative sensor modality for accurately estimating amounts and flow, with a mean RMSE of 2.8g across the five materials for pouring. We also demonstrate how the learned networks can be used to pour a desired amount of material.

BibTeX

@conference{Clarke-2018-110189,
author = {Samuel Clarke and Travers Rhodes and Christopher G. Atkeson and Oliver Kroemer},
title = {Learning Audio Feedback for Estimating Amount and Flow of Granular Material},
booktitle = {Proceedings of Conference on Robot Learning},
year = {2018},
month = {October},
volume = {87},
pages = {529-550},
publisher = {Proceedings of Machine Learning Research},
}