Research On Drivers' Avoidance Behavior In Critical Lateral Conflicts At Intersections

Side-impact accident is the most prevalent form of accident in China,and the driver's factor is the most important causation of traffic accidents.Therefore,it is of great significance to study the behavioral patterns of drivers in lateral conflicts.Based on the data collection of traffic accidents at intersections in Beijing,and the criteria for the longitudinal and lateral emergency avoidance operations of the driver before the accidents,this article analyzed the driver's avoidance behavior before the accidents.Using accident reconstruction method,this dissertation found a clear tendency that drivers usually overuse Same-Direction-Swerving as well as braking in lateral conflicts,which caused many accidents that could have been avoided.Using kinematic analysis methods,through the concept of Priority Level,the optimal vertical,horizontal,and combination avoidance strategies under rational decision-making conditions were analyzed.It is proved that the Same-Direction-Pattern and Braking-Pattern are important causations of the accidents.In order to overcome the sample bias brought about by the actual accident data collection,the above patterns were further validated by questionnaires and driving simulation experiments.Based on the driving simulation experimental platform,the static 3D scenes and dynamic traffic elements for the intersectional driver behavior research were designed and experimental research was conducted.Through systematic parametric tests on the initial conditions of conflicts such as urgency level,initial priority level,obstacle size,obstacle speed,and direction of conflict,the driver's avoidance behavior rules were analyzed.A generalized linear mixed model with cumulative logit as a linkage function was established for the driver's horizontal and vertical operation behavior under different initial conditions of conflicts.Based on the theory of bounded rational decision-making,hypothesis of decision models that can be used to explain the driver's evasive tendencies was proposed.Through the questionnaire testing method,the explicit attitude of the driver on the dimensions of intuition/reasoning,collision avoiding/injury reducing of injury was studied;through the lexical decision task testing,the driver's implicit attitude of intuition/reasoning,collision avoiding/injury reducing under different urgency levels were studied.For the verification of intuition and reasoning decision-making,the conflict conditions that can accurately reflect intuitive decision-making and reasoning decision-making were selected,and the difference in response time between the two was analyzed.Comprehensively,the evaluation indicators found in this paper were used to evaluate the decision hypothesis.Finally,the Keep-Away heuristic dual-system decision model with the strongest explanatory power is selected as the driver's avoidance decision model,and the theoretical basis of the decision model is analyzed from the perspective of ecological rationality.Two external intervention methods-driver training and directional instruction-were conducted trying to correct the drivers' reaction pattern in lateral conflicts at intersections.Drivers' behavior was examined by driving simulator before the intervention,after the intervention,and in the follow-up tracking experiment,in order to test the effect of the external interventions.Based on the driver behavior patterns and kinematics analysis method found in this study,the collision avoidance strategy for intelligent vehicles with/without vehicular ad-hoc network was designed,which can provide supports for driver assist systems and intelligent vehicle design in future.