再帰型ニューラルネットワークを用いた筋電位による床反力推定

抄録

Japan is facing the problem of super aging society and the necessity of assistive system for the elderly has increased. In order to decrease the human resource and cost, we need to develop the technology to predict and support human motion. Conventional motion prediction system uses force plate or motion capture but these system is limited by the environment so we focus on the Electromyography(EMG) which is available as portable sensors. Our motion is generated by the force and the force is generated by our muscle. Therefore, we consider we may predict our motion from muscle activity, which is the origin of our motion.

In this research, as a first step to realize the better motion prediction system, we try to predict ground reaction force(GRF) from EMG using Recurrent Neural Network.