Research On Pedestrian-vehicle Conflict Discrimination And Resolution At Crosswalks For Autonomous Vehicles

Pedestrians are vulnerable road users and are at a greater risk of being injured in traffic crashes all over the world.Especially at uncontrolled mid-block crosswalks,due to the lack of signal control,the phenomenon of pedestrian-vehicle interaction is very complex,and may even occur pedestrian-vehicle collision here.With the development of autonomous driving and other related technologies,drivers can partly get rid of tedious driving and improved driving safety.At present,there are many studies on autonomous vehicles and pedestrian collision avoidance,most of which focus on pedestrian identification and pedestrian-vehicle collision risk prediction.However,research on how to avoid collision between autonomous vehicles and pedestrians,and improve the efficiency of pedestrians and vehicles on the premise of ensuring the safety of pedestrians and vehicles are less.Therefore,it is necessary to analyze the phenomenon of the interaction between pedestrian and vehicle,accurately distinguish pedestrians ’ crossing trend and possible conflict between pedestrians and vehicles.On the premise of guaranteeing the safety of pedestrians and vehicles,an efficient conflict resolution method is proposed.This study can provide support and basis for autonomous vehicle driving at crosswalks.In order to study the safety of pedestrian-vehicle interaction,this paper takes uncontrolled mid-block crosswalk as an experimental scenario.On the basis of considering the characteristics of pedestrian crossing behavior,this research analyses how to accurately distinguish the conflict between pedestrians and vehicles.Combining the current rules of vehicle yielding to pedestrian in China,considering the safety and efficiency of pedestrians and vehicles comprehensively,a reasonable method of conflict resolution is proposed.The main work of this paper is as follows:Firstly,the decision-making of pedestrian crossing at uncontrolled mid-block crosswalks is analyzed.Considering that the gap acceptance behavior of pedestrians will affect their crossing decision-making.Therefore,based on pedestrian crossing data and traffic flow data of nine uncontrolled mid-block crosswalks in Wuhan,a pedestrian crossing decision-making model based on dynamic gap acceptance is established by binary logistic regression.Moreover,the proposed model is validated by the actual data.Secondly,this paper proposes the discrimination method of pedestrian-vehicle conflicts at uncontrolled mid-block crosswalks.Due to different pedestrian crossing decisions will have a greater impact on pedestrian-vehicle conflict.This research establishes a pedestrian-vehicle conflict discrimination model based on pedestrian crossing decision-making.Moreover,this research gives the process of conflict discrimination,and introduces the method of conflict discrimination for autonomous vehicles under different pedestrian crossing decisions in detail.Thirdly,this paper proposes the method of pedestrian-vehicle conflict resolution at uncontrolled mid-block crosswalks.According to the conflict identified in the discriminant model of pedestrian-vehicle conflict,and considering pedestrian safety,pedestrian-vehicle efficiency and following vehicles,a conflict resolution method for autonomous is proposed based on the rules of vehicles yielding to pedestrians.Finally,the pedestrian-vehicle interaction simulation experiment is carried out.Based on the decision-making model of pedestrians crossing and the pedestrian-vehicle conflict resolution model,the traffic rules of autonomous vehicles and pedestrian crossing rules at uncontrolled mid-block crosswalks are given.Using MATLAB simulation software,the cellular automata model is used to conduct the simulation.According to the experimental results,the pedestrian-vehicle safety and efficiency are taken as evaluation indicators to verify the feasibility of the method of conflict resolution.The safety of pedestrian-vehicle interaction is an important research content in road traffic.This paper establishes a pedestrian-vehicle conflict discrimination model based on pedestrian crossing decision-making and a conflict resolution model for autonomous vehicles.The models have referential significance for autonomous vehicles and manual driving vehicles on how to ensure the safety and efficiency of pedestrians and vehicles at crosswalks.