基于车道居中控制的驾驶员切弯行为偏好视觉影响机制

Visual Mechanism Behind Drivers’ Preference on Curve-cutting Behavior of Lane Centering Control during Curve Negotiation

摘要:

驾驶模拟器的采集车道居中控制系统（LCCS）开启时，根据高速公路匝道弯道上的驾驶员注视点视角与关键路点视角信息，建立模型分析视觉特征与该工况下驾驶员对路径切弯行为偏好的关系。首先，提出基于临近路点的驾驶员注视行为分析方法；其次，分析不同弯道区间段上关键路点与驾驶员注视视点之间的位置关系，以及利用其预测偏好的可行性；最后，基于视觉特征设计了72个统计指标，筛选后选择其中8个指标建立切弯行为偏好估计逻辑回归模型。结果表明：该模型能够准确估计驾驶员对两条路径在弯道中心区域切弯程度的相对偏好；模型中的参数反映了切弯行为偏好背后的视觉影响机制，这为弯道居中控制时的驾驶员偏好获取及其自适应方法提供了依据与支撑。

Abstract:

The visual angles of drivers' gaze points and critical waypoints on an expressway ramp in the driving simulator when the lane centering control system （LCCS） is on were gathered. A model was built to analyze the relationship between visual characteristics and drivers’ preference on the curve-cutting behavior of paths. First， methods to analyze driver gaze behavior by nearing critical waypoints were proposed. Next， the relationship between positions of critical waypoints and driver gaze points on different curve segments was analyzed， and the reason for this relationship was used to predict driver preference was also demonstrated. Finally， 72 statistic indices were designed based on visual characteristics. A logistic regression model of drivers’ preference on curve cutting behavior with selected 8 indices was built. The results show that the estimation model predicts drivers’ relative preference on curve cutting behavior of two paths on the center area of the curve road with a high prediction accuracy. Coefficients in the model reveal the visual mechanism behind curve-cutting behavior preference， which provides the basis for the design of preference estimation methods and lane-centering control systems adaptive to driver preference.