Feature Extraction of Pedestrian Behavior Propensity based on BP Neural Network | |
---|---|
( Volume 3 Issue 2,February 2017 ) OPEN ACCESS | |
Author(s): | |
Zhenxue Liu, Yaqi Liu, Haibo Wang, Wei Tian, Xiaoyuan Wang | |
Abstract: | |
Pedestrian is an important part of the traffic system, pedestrian travel safety, efficiency, comfort and so on have received more and more people's attention. Under the premise of the internet of pedestrians, it is of great significance to timely implement pedestrian safety warning for improving its active safety. The prerequisite for the implementation of the safety warning is to accurately identify the pedestrian's intention, and the pedestrian's intention is affected by many factors, among which the difference of the individual characteristics of the pedestrian is an important factor causing the difference of the movement intention. Thus, the reasonable pedestrian classification is of great significance to construct the scientific and reasonable pedestrian safety early warning system. Aimed at the differences of pedestrian traffic microcosmic behavior, the individual characteristics of pedestrians influencing pedestrian movement intention were analyzed thoroughly. Limited to the availability of psychological parameters of pedestrian individual differences and the external influencing factors, the concept of pedestrian behavior propensity was put forward learning from the concept of driving propensity, and it was used to describe the differences of the individual characteristics. Pedestrians were divided into three types of safety, task-based and comfortable according to the difference of actual pedestrian traffic behavior. Combined with the questionnaire survey, the non-invasive natural walking observation experiment was used to collect the movement data of three types of pedestrians. The pedestrian behavior in free flow was taken as an example, the feature vector of pedestrian behavior propensity was analyzed and extracted based on BP neural network. This study can provide theoretical support for the construction of scientific and reasonable pedestrian classification model and personalized pedestrian safety warning system. |
|
Paper Statistics: | Total View : 576 | Downloads : 567 | Page No: 9-14 | |
Cite this Article: | Click here to get all Styles of Citation using DOI of the article. |