Reliable Safety Evaluation Method On Multi-vehicle Highway Of Autonomous Driving

As the increasing development of the transportation and vehicle industry,the contradiction between drivers and road infrastructure has induced amounts of traffic accidents.For this reason,countries around the world have carried out research on autonomous vehicles to reduce the rate of traffic accidents and improve travel efficiency.However,issues such as cost,technique,and road traffic management still restrict the popularization and application of civilian autonomous vehicles.Among them,road is an important part of the human-vehicle-road-cloud system.Because of people’s concern about road safety under autonomous driving,there is a new demand for road design methodology.However,while the existing road network has a higher degree of maturity and complexity,relatively complete road functions,reconstruction need large costs and work quantity,which may cause great wastes.After investigation,this paper proposes a new safety evaluation of existing roads in combination with the driving characteristics of autonomous vehicles,and proposes improvement measures for road designs that do not adapt to new vehicles.Firstly,this article systematically outlines the principles of environmental perception and recognition of autonomous vehicles,explores the system data sources and road detection methods,and analyzes the limitations of autonomous driving technology and possible new traffic accidents.Combined with the technology,the road surface type,traffic facilities,road alignment,traffic flow status and other aspects are considered together to further analyze the differences between the physiological and psychological characteristics of the driver and the automatic driving system,and the reasons that may lead to accidents,as a safety analysis and reliability modeling Prerequisites.Considering the significant difference between ADS and drivers,this paper also analyzes the limitations of traditional driving road safety assessment methods,and revises the existing road indicators such as parking sight distance,auxiliary driver’s workload,cross-sectional width,traffic facilities,etc.One of the measurement standard of automatic driving recognition ability is proposed as the road pollution index.And the correspondence between the pavement skidresistant performance index and the pavement pollution index is verified.Based on the Fault Tree Analysis,the abnormal indicators of possible accidents leading to traffic accidents are comprehensively sorted out,and five scenarios of rollover,side slip,poor line of sight,bumps,and congestion are selected to determine the reliability threshold function of curve radius,line of sight,and acceleration and the impact of the market share of autonomous vehicles on the stability of mixed-road traffic flow.The Monte Carlo method is used to calculate the failure probability under the Confidence Interval of 99%.And a safety evaluation methodology for multi-vehicle highway is established.Finally,based on the data in K55+806～K57+700 and K7+005～K10+650,sections of a provincial highway in Guizhou Province,this methodology is used to evaluate the ability and safety of this highway where autonomous vehicles and conventional vehicles drive on in the same time.Combined with historical accident data analysis,the ability of this methodology is verified.And the suggestions for improvement of transportation facilities in unsuitable highway are proposed.