| In order to improve the driving safety,the intelligentization and networking have become the development trend of vehicles.Vehicle-assisted safety systems based on vehicle-mounted machine vision and radar are easily affected by road environment occlusion.How to achieve vehicle-vehicle collaborative assistance safety through network connection has become one of the current research hotspots with the development of C-V2 X technology.Among the many factors that induce traffic accidents,the reasons caused by drivers account for the main proportion.Therefore,combining driver's abnormal behavior and C-V2 X communication to realize vehicle-vehicle collaborative collision warning can effectively reduce the occurrence of traffic accidents.Through the analysis of the collision warning scene under the vehicle-vehicle collaborative environment and the impact of collision warning because of the driver's abnormal driving behavior,the model of the vehicle-vehicle cooperative collision warning based on the driving behavior is studied and designed in this thesis.It provides support for vehicle cooperative collision warning.The main research work of this article includes several aspects:1.First,a driver reaction time model is established based on driving behavior in this thesis.In the traditional vehicle collision warning technology,the impact of the driver's abnormal driving behavior on the vehicle's braking process is not taken into account,so it cannot adapt well to the driver's different behaviour.The impact of the driver's typical abnormal driving behavior on the vehicle's braking process is studied in this thesis.Through the survival analysis method,it is verified that abnormal driving behavior has a significant effect on the driver's reaction time during emergency braking.Through the driver reaction time model and considering the different motion states of the vehicle,the emergency braking reaction time corresponding to the different abnormal driving behavior is calculated.Through model training and testing,the test accuracy of XGBoostBP combined model is 87.80%.2.Secondly,a vehicle collision warning algorithm based on reaction time is designed in this thesis.According to the driver reaction time,the traditional vehicle safety distance model is improved to the vehicle safety distance model in typical scenes such as straight roads,curves and intersections.Combined with V2 X communication technology,the vehicle collision warning algorithm based on the typical scene safety distance model is designed.In order to improve the accuracy of warning,it can effectively identify different traffic scenarios and use corresponding collision warning algorithms for different traffic scenarios.3.Then,in response to the data parallel processing requirements of OBU device,a layered vehicle collision warning software architecture is designed in this thesis.The vehicle collision warning architecture includes a data acquisition module and a communication module at the bottom,a data cache module at the middle layer,and an vehicle collision warning module at the application layer from the bottom up.The bottom module provides the original vehicle motion data for the middle layer,the middle layer provides data caching,synchronization and retrieval functions for the application layer.The application layer is responsible for the operation calculation of the vehicle collision warning algorithm.Finally,the function test of collision warning application in the situation of straight roads,bends and intersections shows that the result meets the functional requirements of collision warning.The accuracy of the collision warning algorithm is tested,and the warning accuracy rate reaches 81.3%,the false alarm rate is 12.2%.The real-time test of the collision warning application shows that the overall delay of the collision warning system is between 65-85 ms,which basically meets the time delay requirements for auxiliary safety applications in the national standard,and the system has good real-time performance. |
PDF Full Text Request