Home // ACHI 2020, The Thirteenth International Conference on Advances in Computer-Human Interactions // View article


Simple Generative Adversarial Network to Generate Three-axis Time-series Data for Vibrotactile Displays

Authors:
Shotaro Agatsuma
Junya Kurogi
Satoshi Saga
Simona Vasilache
Shin Takahashi

Keywords: Acceleration, Generative Adversarial Networks, Vibrotactile Display.

Abstract:
Various kinds of vibrotactile information have been recorded from real textures and used to present high-quality tactile sensations via tactile displays. However, it is unrealistic to collect large amounts of vibrotactile data under many different conditions. Thus, we develop a method whereby recorded data can be changed to represent conditions differing from those at the time of initial recording. In the first step, we construct a data generation model using a Generative Adversarial Network (GAN). The model makes simple calculations and generates unknown data from recorded acceleration data obtained by rubbing real objects. The model can generate three-axis, time-series data. To evaluate the quality of the data generated, we devised a string-based tactile display and presented generated vibrotactile information to users. Users reported that the generated data were indistinguishable from real data.

Pages: 19 to 24

Copyright: Copyright (c) IARIA, 2020

Publication date: March 22, 2020

Published in: conference

ISSN: 2308-4138

ISBN: 978-1-61208-761-0

Location: Valencia, Spain

Dates: from November 21, 2020 to November 25, 2020