@techreport{oai:ipsj.ixsq.nii.ac.jp:00218326,
 author = {福田, 純 and 福井, 航 and 平田, 一郎 and 後藤, 泰徳 and Atsushi, Fukuda and Wataru, Fukui and Ichiro, Hirata and Yasunori, Goto},
 issue = {1},
 month = {Jun},
 note = {One concrete example of human-centered design (HCD) is design that takes into account ergonomic influences using musculoskeletal models. Methods using motion capture are often used to calculate muscle loading on musculoskeletal models, but these methods cannot be tested in an arbitrary field. Therefore, we thought that if AI-based pose estimation technique adequately combined with a musculoskeletal simulator, this method could be a powerful tool in performing human-centered design. In this study, the results of AI-based pose estimation were used as input to a musculoskeletal simulator to estimate muscle activity. When limited to movements on a plane, the 2D pose estimation confirmed the output as intended. On the other hand, for 3D pose estimation from multiple cameras, keypoint accuracy was insufficient and correct calculation results could not be obtained., One concrete example of human-centered design (HCD) is design that takes into account ergonomic influences using musculoskeletal models. Methods using motion capture are often used to calculate muscle loading on musculoskeletal models, but these methods cannot be tested in an arbitrary field. Therefore, we thought that if AI-based pose estimation technique adequately combined with a musculoskeletal simulator, this method could be a powerful tool in performing human-centered design. In this study, the results of AI-based pose estimation were used as input to a musculoskeletal simulator to estimate muscle activity. When limited to movements on a plane, the 2D pose estimation confirmed the output as intended. On the other hand, for 3D pose estimation from multiple cameras, keypoint accuracy was insufficient and correct calculation results could not be obtained.},
 title = {筋骨格モデリングシミュレータと姿勢推定AI を組み合わせた筋活動量の推定},
 year = {2022}
}