グループ歩行者の存在を考慮した歩行者行動モデル

DOI Web Site Web Site 被引用文献2件 参考文献6件

書誌事項

タイトル別名
  • PEDESTRIAN BEHAVIOR MODEL CONSIDERING THE PRESENCE OF PEDESTRIAN GROUPS
  • グループ ホコウシャ ノ ソンザイ オ コウリョ シタ ホコウシャ コウドウ モデル

この論文をさがす

抄録

<p> In this paper, we constructed a method to detect pedestrian groups using Support Vector Machine (SVM), one of the machine learning methods, from the pedestrian behavior monitoring data measured by laser scanner sensors. In addition, we constructed a pedestrian behavior model taking account the presence of pedestrian groups, and evaluated pedestrian space by simulating the walking behavior as follows.</p><p> (1) First, features that are considered effective for group detection were extracted from the pedestrian trajectory data measured by laser scanner sensors in a hospital, and the relationships between the features and pedestrian groups were analyzed. Furthermore, based on this basic analysis, we constructed a method to detect pedestrian groups who act together as a group from observed multiple walking trajectories. Specifically, five significant parameters for group detection were selected to detect the pedestrian pairs from measured pedestrian trajectory data. SVM was used to detect pedestrian groups, and the effectiveness and efficiency of the proposed method was validated using the pedestrian behavior monitoring data measured by laser scanner sensors.</p><p> (2) Next, the pedestrian behavior model based on psychological stress was extended to a model that can consider the existence of pedestrian groups. More specifically, th e following points were incorporated into the model. (1) The other person stress received from the other pedestrians in the same group is relatively small, (2) The group dispersion stress is caused by leaving from the pedestrians in the same group, (3) The pedestrian stress received from the facing other pedestrian groups is relatively bigger than the stress received from individual pedestrian. Pedestrian groups were detected by SVM, and the parameters of the pedestrian behavior model were estimated for each pedestrian attribute (sex, staff, use of assistive devices). Using the estimated parameters composing each stress, we demonstrated the behavioral characteristics of assistive device users who were difficult to change directions suddenly, and the characteristics of the stress of staff who always corresponded to patients and visitors. In addition, the differences between the estimated trajectory and the measured trajectory were compared, and good description accuracy of the proposed model was confirmed for all attributes.</p><p> (3) Next, the pedestrian space was evaluated using various stress values calculated by the proposed model. The stress received from other pedestrians is larger in places where there are many waiting times, such as reception machines and accounting machines. Also, the group dispersion stress is larger in places where the pedestrian density is low and the distance between pedestrians in the same group is easy to open.</p><p> In conclusion, the pedestrian behavior model based on psychological stress can be of great potential and applied to crowd flow analysis m actual pedestrian spaces. In other facility, such as commercial facilities, however, pedestrian attributes, walking characteristics, and group composition might be different form that of this research. The application of the proposed model in a wide variety of facilities will be discussed in the further study.</p>

収録刊行物

被引用文献 (2)*注記

もっと見る

参考文献 (6)*注記

もっと見る

キーワード

詳細情報 詳細情報について

問題の指摘

ページトップへ