Recurrent neural network for tactile texture recognition using pressure and 6-axis acceleration sensor data

When people decide on preferences of personal belongings, they consider not only the appearance of the object but also the its tactile texture. The design of tactile texture is an important factor in the development of commercial products. To measure a feeling of an object, a numerical evaluation method for tactile texture is required. The measurement is performed using a large-scale apparatus, generally. In this paper, we propose a novel recognition method for tactile texture using long short-term memory recurrent neural network. In proposed framework, the tactile texture information is obtained by analyzing a time-series data of a pressure sensor and 6-axis acceleration sensor. Thus, the system configuration is simple, and it is possible to construct the system inexpensively.