発表論文

2018.01

Predicting Muscle Activity and Joint Angle from Skin Shape

佐川 立昌, 鮎澤 光, 吉安 祐介, Akihiko Murai

概要

Muscle of human body can be a clue to recognize the behavior and intention of a person. If the muscle activity is measured only by visual observation, it is useful to estimate the state of the muscle. In this paper, a method of predicting muscle activity and joint angle of human body from skin shape is proposed. Since the muscle activity and the joint angle affect the skin shape, the both factors should be considered simultaneously. The proposed method is a learning-based approach that uses the data set of the skin shape, the muscle activity and the joint angle. It trains a linear regressor for predicting muscle activity and joint angle from skin shape. The deformation of skin shape is calculated as the feature in the active regions, which are extracted from the training data and limits the regions of the skin shape that contribute to the prediction. We acquired a lower limb with simple motion to consider the small number of factors in this paper. In the experiment, the muscle activity and joint angle are predicted even in the case that the both factors change simultaneously. The skin regions that contributes to prediction are given as the result of learning, and the distribution is reasonable from the viewpoint of biomechanics.