@article{oai:niit.repo.nii.ac.jp:00000011,
 author = {村上, 肇},
 journal = {新潟工科大学研究紀要},
 month = {Dec},
 note = {Functional Electrical Stimulation (FES) is a technique for restoration of lost motor functions of paralyzed patients. Electrical pulses generated by a stimulator are conducted to the neuromuscular system of a patient. His/her muscles are activated by the pulses, and a desired motion is elicited. Stimulus data (a set of intensity of the pulses for each muscle) for the desired motion have been created and stored in a stimulator. In the current FES system, they were created from electromyogram (EMG) signals which were measured from healthy subjects. Although this method can generate stimulus data whlch reflect natural activities of the normal muscle system, it requires complicated procedure for EMG analysis. In this paper, a creation method of the stimulus data by using Artificial Neural Network (ANN) which mimics the musculoskeletal system is reported. Direct inverse modeling which is adopted as a learning method of ANN cannot be applied to a redundant object such as the musculoskeletal system. Hence a constraint is used for the resolution of kinematic redundancy. The method is verified through simulation study with a wrist joint simulator which has four muscles (synergists and/or antagonists) and elicits ulnar/radial flexion and/or palmar/dorsi flexion.},
 pages = {29--34},
 title = {人工神経回路を用いた運動機能の制御 : シミュレーション解析},
 volume = {1},
 year = {1996}
}