Hand-Motion Intention Recognition Based on One-Dimensional Convolutional Neural Network

7Surface electromyographic (sEMG) signals play an important role in human-computer interaction between patients and rehabilitation robots in rehabilitation training. This paper shows the effectiveness of one-dimensional convolutional neural networks (1-D CNN) in estimating hand motion intentions from sEMG data. Ninapro database 1 was employed to train the model. First, the data from a single subject was used to train a model, and the accuracy was roughly maintained at 80%. Then, the whole dataset was used to train the network, and the recognition accuracy was improved to 84.16%. This shows that 1-D CNN effectively estimates hand movements and achieves high accuracy even for a small number of data.