Vibrotactile Signal Generation by Deep Convolutional Generative Adversarial Network for Vibrotactile Displays

Abstract

To create intuitive tactile displays, collecting vibrotactile information is important, though, the collection procedure requires manual scanning of textures. Thus, a collection of vast information is difficult. However, by employing machine learning technology, there is a possibility to generate further virtual data from existing collected data. In this paper, we made a generation model of vibrotactile signals from the collected acceleration data by using Deep Convolutional Generative Adversarial Network (DCGAN). We held two kinds of experiments to evaluate the performance of our DCGAN model and we consider the result of experiments.