Research On Risk Level Evaluation Of Driving Environment Of Smart Cars Based On Bayesian Network

According to the development process of car intelligence and automation,the Society of Automotive Engineers has divided the development of smart cars into level 0 to level 5.Although smart cars of different levels and functions are developing rapidly,it is very difficult to achieve true fully automatic driving in the short term.Therefore,for a long time in the future,smart cars will inevitably face the situation of human-machine cooperative driving.However,the challenges that need to be solved urgently in human-machine cooperative driving are the uncertainties: uncertainty of the driver,uncertainty of the vehicle,and uncertainty of whether the driving right is conditionally switched,and so on.To solve the above uncertain issues,this paper establishes a driving environment risk level assessment model based on the Bayesian network.The model combines driver factors,vehicle factors,and external environmental factors,which can quantitatively output the risk level of the driving environment.First ly,the factors affecting driving risk are divided into two main aspects: original risk and dynamic risk.Among them,the dynamic risk is divided into driver dynamic risk and vehicle dynamic risk,and the relevant impact indicators are analyzed and selected;Secondly,according to various influencing factors,a Bayesian network of driving environment risk is built,and a driving environment risk level model is derived through the network;What’s more,to analyze and optimize vehicle driving risk indicators,and delineate the risk level range of related impact indicators;Finally,Netica is used to conduct sensitivity analysis and risk prediction on the driving environment risk level model,and c ombined with relevant driving data to prove the feasibility of the model.Besides,the effectiveness of the model is verified through two actual vehicle application scenarios.The research results of this article can provide theoretical support for L3 and L4human-machine cooperative driving cars when the driving rights are switched.When it is necessary to switch the driving rights,the system outputs the risk of the driver and the vehicle through the model to determine whether the current conditions for the transfer of the driving rights are currently available,to decide whether to initiate the transfer of the driving rights;besides,the model can also monitor the overall risk in the driving process quantitatively and in real-time,to correct the driver’s dangerous driving behavior at any time.