不同干预方式下右转车辆避让行人效果研究

为探究车辆右转过程中不同干预方式对驾驶人未规避行人行为的改善情况，设计听觉警示、触觉警示、形式惩罚、利益惩罚和道德惩罚5种干预方式，分为控制组、警示组和惩罚组，试验基于眼动仪和模拟驾驶仪展开。定义注视次数、注视点分布信息熵、平均注视时间、视线转移路径、区域关注概率和瞳孔面积6项指标表征驾驶人眼动特性，提取制动踏板深度比例、行车速度2项指标反映车辆运行状态。经方差分析确定各干预方式差异的显著性水平，从注视特性指标、扫视特性指标、瞳孔面积指标、驾驶人制动指标和机动车制动指标5个方面分析不同干预下驾驶人视觉及操纵响应特征，并收集被试反馈的追踪问卷。试验结果表明：不同干预方式对右转车辆未避让行人均有规范作用，各组干预效果由强到弱依次为利益惩罚、道德惩罚、形式惩罚、触觉警示和听觉警示。利益惩罚性主动干预效果优势显著，注视点分布信息熵最高为0.74，右侧平均注视次数为6次，平均注视时间增加至13.25 s，驾驶人对右侧注视概率增加至0.403，瞳孔面积明显增大，制动踏板深度比例维持在0.8，右转车速下降至20 km·h^-1以下，谨慎驾驶程度和避让行人意识均有提升。一致性追踪问卷调查表明，结束试验时32%的驾驶人对利益惩罚印象深刻，驾驶人对其主观认可度高达83%，具有较强的推广性；该干预方式可帮助驾驶人规范驾驶行为，树立避让行人的安全驾驶意识。

Effect of Intervention Mode on Avoidance of Pedestrians by Right-turning Vehicles

To study the improvement in the avoidance of pedestrians by drivers while turning right, five intervention methods were designed, namely, auditory warning, tactile warning, formal punishment, interest penalty, and moral punishment. These methods were categorized into the following groups, namely, control group, warning group, and punishment group for the simulation of a driving test with the eye tracker and simulator. The number of gazes, gaze distribution information entropy, average gaze time, line of sight transfer path, regional attention probability, and pupil area were six indicators considered for the characterization of a driver's eye movement. The brake depth and driving speed were extracted to characterize the state of the vehicle. The significance level of each intervention method was determined via analysis of variance. The visual and behavioral responses of the drivers under different interventions were analyzed with respect to five aspects:gaze characteristic index, saccade characteristic index, pupil area index, driver braking index, and motor vehicle braking index. The subjects were provided with a tracking questionnaire to obtain feedback on the methods. The test results demonstrate that the five intervention methods have a normative effect on the avoidance of pedestrians by drivers when turning right. The effects of the intervention methods can be arranged from strong to weak as interest penalty, moral punishment, form punishment, tactile warning, and auditory warning. The beneficial effect of the punitive active intervention is significant; after receiving the penalty, the driver possesses the highest information entropy of 0.74 and the average number of gazes to the right is 6, which is greater than that for the other groups. The average gaze time increases to 13.25 s with a statistically significant difference. After accepting the punitive intervention of interest, the driver's right gaze probability increases to 0.403, pupil area increases significantly, brake pedal depth ratio remains at 0.8, and right-turn speed decreases to below 20 km·h^-1; that is, the driver becomes more cautious while driving and has a higher awareness with respect to avoiding pedestrians. The consistency tracking questionnaire results reveal that 32% of the subjects are impressed with the penalty of interest and the driver's subjective recognition is as high as 83%. The interest penalty has strong promotion and help drivers standardize driving behavior, and establish a sense of safe driving to avoid pedestrians.