| Red-light-running(RLR)behavior is an important factor that affects intersection safety and usually leads to severe crash consequences.The rapid development of road-vehicle coordination technology provides a new direction for the intelligent warning of RLR behavior and crash.How to design a reasonable and efficient intelligent collision warning plan to assist and improve driver behavior and reduce accident rate is the key problem to be solved in this study.Thus,this paper makes a comprehensive research on the characteristics of RLR behaviors and crashes,and designs a reasonable and efficient intelligent collision warning scheme to explore the influences of the intelligent warning technology on the driving behavior.The results could provide scientific theoretical basis for the development of the technology.The main research content includes five aspects as shown in the followings:(1)Analyze the RLR behavior feature and risk by using the RLR data in GES databaseThe RLR behavior was divided into straight-forward RLR and left-turn RLR according to the RLR drivers’ pre-crash behavior.Quasi-Induced Exposure Method was used to analyze the RLR behavior feature and related risk based on this division.Results of the study indicated that:(ⅰ)drivers with age lower than 20 or higher than 60 year-old had a high RLR risk;(ⅱ)drivers driving in dusk or dawn had a high RLR risk while in night driving without light,the RLR risk was highest;(ⅲ)driver distraction,low visibility environment and speeding could all increase the straight-forward and left-turn RLR risk;(ⅳ)drivers aging over 50 year-old,distracted drivers,or drivers driving in low visibility environment or at night had a higher risk in left-turn RLR than in straight-forward RLR;(ⅴ)speeding drivers had a higher risk in straight-forward RLR than in left-turn RLR.(2)Classify the RLR crashes by using the GES RLR crash database and compare the features of different crash typeBased on the responsibility certification of drivers in RLR crash and the pre-crash behavior,the study divided the RLR crashes into three types.Considering the direction where vehicles approached,the three types were further divided into eight subclasses.Taking into account the vehicle,road and environmental factors,the characteristics of the sub-types of RLR crashes for each type were compared and analyzed using a classification and regression tree model.Resutls indicated that for straight-forward RLR vehicles,as the speed limit increased,the crash rate with straight-forward non-RLR vehicle from the right adjacent lane was almost equal to crash rate with vehicle from the left adjacent lane,and meanwhile the crash rate with left-turn non-RLR vehicle from opposite lane increased;for left-turn RLR vehicles,as the number of lane and speed limit increased,crashes between left-turn RLR vehicle and straight-forward non-RLR vehicle from the opposite lane showed an increasing tendency.(3)With consideration of different driving environment,investigate the drivers’ driving behavior and eye-movement when the traffic light turned into yellow and test the effectiveness of RLR warning technologyThe study designed four typical driving scenarios(driving in clear daytime,distracted driving in daytime,driving in heavy fog and driving at night time),in which drivers encountered the traffic light changing from green to yellow.A two-stage RLR warning technology was achieved by using the driving simulator platform.Results of binary logistic regression model showed that distracted drivers,or or drivers driving in fog or night time had a high RLR rate.The linear mixed effects model was used to analyze the driver’s driving behavior and eye movements.Behavioral compensation was observed in different driving environment,including the initial speed when yellow light was triggered,the maximum deceleration,the fixation duration on signal light and the number of fixations on signal light.Besides,the two-stage RLR warning information helped in reducing the RLR rate and prompted drivers to take a smooth deceleration.Furthermore,the warning information showed different effectiveness under different driving environment.The warning information had a largest impact on the reduction of drivers’ speed at the onset of yellow,brake reaction time to yellow light and maximum deceleration rate in heavy fog,then followed by the night time driving and distracted driving environment.Beside,a two-fluid model was used to perform a driver’s RLR risk assessment.It was found that the driver’s behavior was conservative in adverse driving environments,and driver’s individual differences were minimal in heavy fog,and the individual differences in risk acceptance of drivers were greater in nighttime driving.Finally,according to the results of driver’s subjective survey questionnaire,it was found that the two-stage RLR warning information is the most helpful when driving in heavy fog.(4)Design a collision avoidance scenario with straight-forward RLR vehicle from the right adjacent lane,and investigate the effects of different warning releasing time and warning content on drivers’ collision avoidance behavior and eye movementThe study designed 7 types of warning releasing time(2,5s to 5.5s)and 2 types of warning content(with/without direction information).The linear mixed effects model was used to analyze the driver’s driving behavior and eye movements.Results showed that the warning information could reduce crash rate,shorten the drivers’ brake reaction time and improve drivers’ abilities in taking evasive actions.As the warning information was released more ahead,the collision rate and collision possibility with RLR vehicle reduced gradually.When the warning was released 4.5s in advance,drivers could notice the RLR vehicle most quickly and take active saccadic eye-movement on the RLR vehicle.The direction information involved in the warning could assist drivers to locate the potential hazard target timely,increase drivers’ saccade and supervision on the RLR vehicle and thus reduce drivers’ brake reaction time.The results of driver’s subjective survey questionnaire showed that 67.4%of the drivers regarded the directional warning information was more useful.(5)Considering the direction where RLR approached and the driving environment,investigate the drivers’ driving behavior and eye movement in the collision avoidance with RLR vehicle and test the effectiveness of RLR collision avoidance warning technologyThe study designed five RLR types and four types of driving environment.The linear mixed effects model was used to analyze the driver’s driving behavior and eye movements.Results showed that drivers’ abilities to perceive potential hazard target degraded when driving in distraction or heavy fog.In night time driving,drivers were more active in saccadic behavior to observe the potential hazard target.In terms of different RLR approaching directions,the warning information had greatest effect on the collision rate with left-turn RLR vehicle from opposite direction;drivers had more obvious compensation on the deceleration rate in collision avoidance with straight-forward RLR vehicle and left-turn RLR vehicle from opposite direction.For different driving environment,the warning information led to a largest decrement of crash rate in heavy fog condition and a smaller decrement in daytime distracted driving.From the perspective of subjective perception,drivers believed that the RLR vehicles from opposite direction were most dangerous;the warning helped most in heavy fog driving;and 86.9%of the drivers would like to apply the RLR collision warning system in vehicle. |
PDF Full Text Request