Handy-Type Tactile Sensor for Object Recognition Using Convolutional Neural Networks

Tactile sensation obtained from touch is one of the most important factors that determine impression an object. A method for identifying tactile textures required because individual differences exist in feeling textures. In this study, we propose a handy-type sensor object recognition using convolutional neural networks (CNNs). The consists three-axis pressure and optical motion mouse can detect time-series data. identified CNN data, namely, speed sensor, when moved by user hand. Thus, system configuration simple without needing drive device, it be possibly constructed at low cost. Fifteen types objects were prototype sensor. total average correct rate specific study was 77%. Further, four separate users considering each use 48%. Although problem identification remained, result demonstrated potential application. proposed used as functional useful device.