抄録
We propose a new method for joint angle and torque estimation by using ultrasound. The ultrasound is emitted from the surface of skin. The estimation is carried out based on the feature value extracted from the reflected wave from inside of living body. The feature value changes as the movement of the muscle relating to the joint. We conducted an examination of simultaneous joint torque and angle estimation in dynamic motion of human elbow joint. Neural network and Multiple linear regression was employed as the estimator. Correlation coefficients and Mean absolute error between estimated data and measured data were calculated. The results show that the estimation was successful, though the error increased with time course.
