摘要
城市交通枢纽是居民出行的重要支撑节点,普遍具有客流密集、空间封闭、多方式客流复杂交织等特征,也承担了巨大的安全组织与运维的压力。因此,对枢纽内行人异常行为的管控是保证其安全运维的重要关注点之一。目前,对枢纽行人异常行为的辨识主要是依托计算机视觉技术来实现,即:一种是主要关注客流轨迹和密度差异,利用目标检测技术提取枢纽内行人目标并进行轨迹预测,以分析个体或小群体的异常轨迹数据来判断行人的异常行为;另外一种是利用姿势估计技术提取行人的微观动作数据,通过分析行人的骨骼关节点数据来判断其异常动作和行为意图。故梳理了枢纽内行人行为的研究成果,概括分析了基于模型驱动和数据驱动的行人异常行为研究与应用成果,并解析行人异常行为的致因机理,总结了枢纽面对行人异常行为的管理措施,最后进行了枢纽行人异常行为研究的分析和展望。该综述可以加深对交通枢纽行人异常行为的理解,为枢纽内行人的安全运维提供参考和支持。
Urban transportation hubs are pivotal nodes supporting residents’travel,characterized by high passenger flow,spatial confinement,and complex intertwining of multiple modes of passenger flow,which impose significant pressure on safety organization and operation.Therefore,the controlling abnormal behaviors of pedestrian within hubs is one of the key focal points to ensure the safe operation.Currently,the identification of abnormal behaviors mainly relies on computer vision technology.Namely,one approach focuses on tracking passenger trajectories and density disparities,extracts pedestrian targets within hubs and predict the trajectories by target detection technology and determines the abnormal behaviors by analyzing individual or small group abnormal trajectory data.Another approach involves extracting micro-motion data of pedestrians by pose estimation technology,identify the abnormal actions and behavioral intentions by analyzing pedestrian skeletal joint data.Thus,this review summarizes research achievements,comprehensively analyzes model-driven and data-driven research and application outcomes,elucidates the causal mechanisms and summarizes management measures of pedestrian abnormal behaviors.Finally,an analysis and outlook on the research of pedestrian abnormal behaviors within hubs are conducted.This review deepens understanding of pedestrian abnormal behaviors at transportation hubs and provides reference and support for the safe operation and maintenance.
作者
赵霞
高源
李之红
郄堃
唐嘉立
Zhao Xia;Gao Yuan;Li Zhihong;Qie Kun;Tang Jiali(School of Civil and Transportation Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Jiangxi Traffic Monitoring Command Center,Nanchang 330036,China)
出处
《市政技术》
2024年第7期124-132,共9页
Journal of Municipal Technology
关键词
交通枢纽
行人异常行为
目标检测
模型驱动
数据驱动
transportation hub
abnormal pedestrian behavior
object detection
model driven
data driven
