Clustering and deduction of typical dangerous scenarios between passenger vehicles and two-wheelers at crossroads

Abstract

In order to provide an effective scenario deduction and construction scheme for research and development of intelligent vehiclesꎬ statistical analysis was conducted on data of accident scenarios between passenger vehicles and two ̄wheelers at crossroads from National Automobile Accident In ̄depth Investigation System ( NAIS) databaseꎬ and two types of basic scenarios with high proportion were obtained. Thenꎬ seven scenario correlation variables were selected for cluster analysis on static characteristics of these two types of scenarios. Finallyꎬ a kinematics deduction model was establishedꎬ and furthermoreꎬ considering actual threshold of dynamic parameters of five types of typical dangerous scenariosꎬ speed ̄distance danger models of them were constructed. The results show that through methods of statistical classificationꎬ cluster analysis and kinematic model superposition deductionꎬ five kinds of 第 4 期 周华等: 十字路口乘用车与二轮车典型危险场景聚类及推演 typical hazard scenes of passenger cars and two ̄wheeled vehicles at intersections can be obtained in conformity with traffic situation in Chinaꎬ and moreover five sets of dangerous scenarios are obtained.