Embedded Identification of Surface Based on Multirate Sensor Fusion With Deep Neural Network

In this letter, we propose a multivariate time-series classification system that fuses multirate sensor measurements within the latent space of deep neural network. our network, identifies surface category based on audio and inertial generated from impact, each which has different sampling rate resolution in nature. We investigate feasibility categorizing ten everyday surfaces using proposed convolutional is trained an end-to-end manner. To validate approach, developed embedded collected 60 000 data samples under variety conditions. The experimental results obtained exhibit test accuracy for blind dataset 93%, taking less than 300 ms machine environment. conclude letter with discussion future direction research.