Estimation of Forearm Motion Based on EMG Using Quaternion Neural Network

抄録

Quaternions are useful for representing data in three-dimensional space, and the quaternion neural network is effective for learning data in this context. On the other hand, estimating biological motion based on myopotential can be performed directly using electromyogram (EMG) signals as the computer interface. The trajectory of human forearm movement within the three-dimensional space can provide important information. In this study, the relationship between the myopotential of the upper arm muscles and the forearm motion was estimated and investigated using a quaternion neural network.