条件自动驾驶接管与高速公路出口小半径匝道平面线形匹配性研究

Compatibility Between the Takeover of a Conditional Automated Vehicle and the Horizontal Alignment of a Small-radius Off-ramp from a Highway

为了解自动驾驶接管后驾驶人行为与现有高速公路出口小半径匝道平面线形的匹配性,基于驾驶模拟仿真平台构建高速公路出口接管场景并进行驾驶模拟试验。以驾驶模式(手动驾驶,接管)、匝道圆曲线半径(50、70、90 m)与匝道交通流密度(每车道5、45 pcu·km^(-1))作为自变量,构建12种驾驶模拟场景。招募30名被试进行驾驶模拟试验,记录试验过程中驾驶人的眼动特征及车辆运动状态。选取瞳孔直径、水平视角分布标准差、垂向视角分布标准差、平均纵向速度变化率、横向偏移量标准差、最小碰撞时间(TTC)共6项指标,分析检验不同因素对各指标影响的显著性。依据检验结果选取指标构建匹配性模糊综合评价模型,对不同匝道半径与交通流密度组合下的接管匹配性进行评估。研究结果表明:相比手动驾驶场景,接管场景中驾驶人瞳孔直径变大,视点水平方向离散程度增加,车辆纵、横向行驶稳定性降低;随匝道半径增加、交通流密度减小,驾驶人瞳孔直径减小,车辆纵向行驶稳定性增加,匝道半径的变化对车辆运动状态的影响比对驾驶人视觉行为的影响更大。在交通流密度为每车道45 pcu·km^(-1)的场景中,增加匝道半径对接管匹配性的提升存在边际递减效应,且较小匝道半径与较高交通流密度的组合会放大各自对接管匹配性的不利影响。当交通流密度较低(每车道5 pcu·km^(-1))时,即使匝道半径取设计极限值,接管匹配性也可达到“高”;而当交通流密度较高(每车道45 pcu·km^(-1))时,匝道半径需接近目前实际工程应用中的最高标准,方可使接管匹配性达到“高”。基于接管匹配性分级,从人、车、路多角度进行考虑提出接管策略优化建议,以提升驾驶人的接管绩效。研究结果对于提升高速公路出口段自动驾驶车辆接管过程的安全性具有指导作用。

This study aims to examine the compatibility between the driver's behavior after a takeover in automated driving and the horizontal geometric design of a small-radius off-ramp from a highway.The takeover scenarios at the highway exit were constructed based on a driving-simulation platform,and a driving-simulation experiment was conducted.The independent variables of the experiment were the driving mode(manual driving,takeover),the horizontal-curve radius of the ramp(50 m,70 m,90 m),and the traffic density(5 pcu·km^(-1) per lane,45 pcu·km^(-1) per lane).Twelve scenarios were constructed.Thirty people participated in the test,and their gaze behavior and the vehicle's motion state during the test were recorded.The pupil diameter,speed,standard deviations of the gaze yaw,gaze pitch,and lane position,and minimum time to collision(TTC)were selected,and the effects of various factors on each indicator were examined.Finally,based on the test results,the corresponding indicators were selected to build a comprehensive fuzzy compatibility model.The takeover compatibility was estimated under a combination of different ramp horizontal-curve radii and traffic densities.The results show that,compared with manual driving,the pupil diameter and gaze-yaw deviation of drivers increase,and the longitudinal and lateral stabilities of the vehicle are impaired in takeover scenarios.As the curve radius increases or the traffic density decreases,the pupil diameter of the driver decreases,and the longitudinal stability of the vehicle is enhanced.The change in the horizontal-curve radius has a greater impact on the vehicle's motion state than the driver's gaze behavior.In scenarios with a traffic density of 45 pcu·km^(-1) per lane,the ramp horizontal-curve radius has a marginally diminishing effect on the takeover compatibility.The combination of a low ramp radius and high traffic density amplifies the adverse effects on the takeover compatibility.With a low traffic density(5 pcu·km^(-1) per lane),the compatibility can reach“high”,even if the ramp radius takes the limit value.With high traffic density(45 pcu·km^(-1) per lane),the compatibility can reach“high”only when the ramp radius approximates the highest standard in current practical-engineering applications.Based on the takeover-compatibility levels,various approaches from the perspectives of humans,vehicles,and infrastructure can be adopted to improve drivers'takeover performance.The results of this study provide guidance for enhancing the safety of takeovers at highway exits.