超音波振動子を用いた触感呈示におけるCGANによる振動情報変換モデルのクラスタ分類効果

抄録

Demand for technology to reproduce tactile sensations is increasing in various fields. In order to reproduce tactile sensation, it is necessary to quantify and analyze the tactile sensation of an object and transform it into data suitable for input to a tactile display. We developed a model using Conditional Generative Adversarial Nets (CGAN) to generate an input signal to a tactile display using an ultrasonic transducer, based on data quantified by tactile sensor and human tactile evaluation. In addition, we examined the effect of cluster classification on the input data to the model. As a result, the reproduction of tactile sensation was improved in the model with cluster classification. These results suggest the effectiveness of a vibration data transformation model with cluster classification.