@techreport{oai:ipsj.ixsq.nii.ac.jp:00211274,
 author = {高橋, 千彩季 and 森野, 博章},
 issue = {28},
 month = {May},
 note = {地震などの災害時に屋内の避難者の避難経路をスマートフォンに記録できれば，要救助者の救助活動の支援に役立つとして，PDR (Pedestrian Dead Reckoning) に着目した研究が行われている．本稿では対象とするユーザの歩行動作の教師データを日常のなかで事前に取得し，避難時の移動状態の推定に用いる手法を提案する．特徴量は回転ベクトルセンサで測定した足の腿上げ角度，2 歩区間の時間長，加速度を用いる．同じ移動状態でも被験者によって判定に用いるべき特徴量が異なることが予備実験で明らかになったため，これら複数のセンサ値による判定結果の多数決をとり最終的な認識結果とした．平地の歩行，走行，階段の昇降の各移動状態において再現率は 8 割以上となった．, Research focusing on PDR (Pedestrian Dead Reckoning) is being conducted as it would be useful to support rescue activities for an injured people if the evacuation route of indoor evacuees can be recorded on a smartphone after a disaster such as an earthquake. In this paper, we propose a method to acquire the teacher data of the walking motion of the target user in advance in daily life and use it to estimate the moving state at the time of evacuation. For the features, the angle of the thigh detected by a rotation vector sensor, the time length of two steps, and acceleration values are used. Since it was clarified in the preliminary experiment that the features to be used for the judgment differ depending on the subject even in the same moving state, the majority of the judgment results based on these multiple sensor values were taken and used as the final recognition result. A recall rate was over 80% in each movement state of walking, running, descending stairs, and ascending stairs.},
 title = {屋内PDRにおける教師データの部分的導入による階段昇降の推定},
 year = {2021}
}