Development of EMG-wrist angle model based on Markov process toward user’s voluntary operation of myoelectric hand

In this research, we aim to develop a stochastic model representing the relationship between surface myoelectric potential and wrist joint angle toward realization of user’s voluntary control of myoelectric artificial hand. For the purpose, an experimental equipment is prepared, and data acquisition is performed for system identification of the proposed model. The myoelectric potential is measured at the time of palmar flexion-dorsiflexion movement of the wrist, and the signal processing of the acquired myoelectric potential and joint probability density function between movements are derived. As a result, we derive a Gaussian mixture distribution from the joint probability density function and construct a model of motion discrimination at the wrist joint by Markov process.