高速公路特长隧道驾驶人视觉特征变化规律及安全状态判别模型研究

Research on Driver Visual Characteristics and Safe State Discriminate Model on Highway Super Long Tunnel

为研究高速公路特长隧道下驾驶人视觉特征变化规律及其安全状态判别模型,要求被试驾驶人进行特长隧道环境的实车实验,同时利用眼动追踪装置采集驾驶人视觉特征参数。在统计分析数据的基础上,建立基于C4.5决策树的单一指标和综合指标判别模型。研究结果表明:相对于非隧道路段,隧道路段驾驶人瞳孔面积更大,注视时间更长;相对于入口段和出口段,行车段驾驶人瞳孔面积更大,而出口段则以短时注视为主;相对于出口段,入口段驾驶人瞳孔面积变化更大,注视持续时间更长;使用单一视觉参数指标可明显区分行车段和非隧道路段驾驶人安全状态,而基于双指标的识别模型可明显区分各路段驾驶人安全状态。研究进一步深化了高速公路特长隧道下驾驶人信息感知模型研究,为建立实时视觉特征判别安全状态提供理论基础。

In order to study driver visual characteristics and safe state discriminate model in super long tunnel on expressways, the real-vehicle experiment with participants under super long tunnel are conducted while driver eyes movement are collected by eye tracker. Based on the statistic data, C4.5 decision tree is used to establish discriminate model of single index and comprehensive index. The results show that driver＇s pupil area is larger and fixation duration is longer under tunnel environment than non-tunnel case. Driver has a larger pupil area in the medium section compared to the entrance and exit while exit is giving priority to short fixation. Driver＇s pupil area is larger and fixation duration is longer under entrance than exit. Besides, single visual parameter can discriminate safe state under medium section and non-tunnel section, while discriminate model based on two indexes can judge each section. This research makes driver information perception model study moving forward under the tunnel environment, and provides theoretical basis for real-time safe state discriminate model.