@techreport{oai:ipsj.ixsq.nii.ac.jp:00218091,
 author = {矢田, 悠人 and 道田, 峻佑 and 小宮, 聖司 and 脇田, 敏裕 and Yuto, Yada and Shunsuke, Michita and Seiji, Komiya and Toshihiro, Wakita},
 issue = {43},
 month = {May},
 note = {自律移動モビリティは歩行者との混在環境を走行することから歩行者との接触を回避する経路生成機能が必要となる．本研究では歩行者の速度を推定したうえで，画像認識を用いて歩行者の顔向きを認識することにより，モビリティに気づいている歩行者（前向き歩行）に対しては最小限の回避により円滑なすれ違いを行い，歩きながらスマートフォンを注視しているなどでモビリティに気づいていない歩行者（下向き歩行）に対しては大きく回避することにより安全なすれ違いを実現する経路生成を提案した．距離，速度，回避の円滑性，安心感の 4 項目について官能評価実験を行った結果，前向き歩行の場合も下向き歩行の場合も，全項目で評価が向上した．特に下向き歩行の場合，距離に対する評価が大きく向上し，歩きスマホなどでモビリティに気づいていない歩行者に対して安全に回避することができた．, Autonomous mobility vehicles travel in mixed environments with pedestrians, so a path generation function is required to avoid contact with pedestrians. This research estimates the speed of pedestrians and recognizes the direction of their faces using image recognition, so that pedestrians who are aware of mobility (forward walking) can pass each other smoothly with minimal avoidance, and pedestrians who are unaware of mobility (downward walking), such as those who are walking on their phones, can pass each other smoothly with minimal avoidance. Path generation that achieves safe passing each other by largely avoiding each other was proposed. Sensory evaluation experiments were conducted on four items: distance, speed, smoothness of avoidance and sense of security, and the results showed that the evaluation improved for all items except speed, both for forward and downward walking. In particular, when walking downwards, the evaluation of the distance was greatly improved, allowing safe avoidance of pedestrians who were unaware of mobility, such as those walking on their phones.},
 title = {歩行者と協調的な自律移動モビリティ-歩行者の顔向きに適応した経路生成-},
 year = {2022}
}