高度自动化驾驶中驾驶人接管车辆时视觉聚焦点转移机理及注意力唤醒人机交互方法

Highly automated driving may cause the driver's distraction as in the process of recovery of vehicle control, which may lead to traffic accidents because of chaos and errors in driver's operation. Therefore, by exploring the mechanism of visual focus shift during takeover of vehicle and putting forward the method of arousing attention, it will play an important role in recontrolling the highly automated vehicle. Through literature review, the shift of focus of attention stimulated by warning signals awakens drivers’ awareness of the current driving situation, thus affecting driving behavior and ultimately driving safety. Firstly, this project will explore the change rule of driver's attention focus shift under warning by regression of the vehicle experiment, and deduce the coupling model of driver's scene awareness cognition and driving behavior and analyze its change characteristics. Secondly, establish the assessment model of taking-over failure degree based on driver's attention focus shift. Moreover, based on the above works, this project is to propose the interactive alarm design method and restoring driving safety measures for driver's scenario recognition and compliance with driving behavior characteristics and optimal safety and efficiency by genetic algorithm, ergonomics and other methods. The research findings have important theoretical significance and practical value for accident causes and prevention when the focus of attention shifts in highly automated driving, as well as for putting forward scientific driving right switching recovery driving control methods.

高度自动化驾驶中驾驶人易分心导致接管时未能将视觉聚焦点转移到当前车况，从而出现操作混乱和错误引发交通事故。因此，通过探究车辆接管时视觉聚焦点转移机理并提出唤醒注意力方法将对恢复行车安全起到重要作用。根据认知心理学相关理论，本项目以警示信号唤醒驾驶人注意力，引起视觉聚焦点产生转移，获取其对当前行车情景的情境意识再认知，进而决策并接管车辆作为假设条件，开展实车实验回归获得警示作用下驾驶人视觉聚焦点转移变化规律，推导驾驶人情境意识认知和驾驶行为的耦合模型并分析其变化特征；建立基于视觉聚焦点转移下驾驶行为的接管失效度评估模型，在此基础上结合人类工效学设计原理，提出增强驾驶人情境认知和符合驾驶行为特性及安全与效率最优的人机交互警报设计方法和恢复行车安全措施。研究成果对高度自动化驾驶中视觉聚焦点转移时事故原因和事故预防，以及提出科学的驾驶人接管车辆的注意力唤醒人机交互方法具有重要的理论意义与实际价值。

目前高度自动化车辆驾驶接管过程中驾驶人接收听觉和视觉警报过程中注意力分配和视觉策略是解决驾驶人从非驾驶任务切换到手动驾驶分心问题的关键。本项目基于信任-情绪-注意力转移理论，研究了基于聚类群体的信号-情绪-注意力转移路径，通过自动化车辆实车实验测量了驾驶员的接管反应，揭示了基于不同等级听觉警报系统的驾驶人在自动驾驶接管的响应机制，即符合警报响应的大脑信息资源分配法则，在双任务的范式下驾驶人的响应遵循最短路线原则。实验同时揭示了不同等级听觉警报系统的信任度的驾驶人响应概率匹配规律，反映了自动驾驶车辆接管过程中驾驶人信任、情绪、等级警报、接管绩效之间的相互耦合关系。为解决高度自动驾驶接管驾驶员分心问题和提高接管绩效提供了理论依据和实际价值。主要工作和成果综述如下：. 1、基于高度自动化车辆驾驶人认知负荷分析和态势感知原理，构建影响高度自动化车辆驾驶人接管驾驶的结构方程，探究驾驶人接管自动化车辆的认知-情绪-注意力的耦合关系和影响因素。. 2、从不同任务承担角度对高度自动化车辆驾驶接管全过程的驾驶人进行聚类分析，探究了司机角色和监管者角色在自动化驾驶接管全过程中的不同表现。监管者角色时注意力集中在非驾驶任务，情绪变化不明显；司机角色时人类防御机制被激活，注意力集中在驾驶中，自主意识强烈。.3、采用语音、标准音和听觉图标三种警报方式研究了高度自动化驾驶接管中驾驶人的表现，视觉策略采用是环顾四周各个方向，他们将注意力放在前车道路，并且检查仪表盘，然后检查左边的后视镜，以及右侧车道和右边的后视镜。.4、通过实验研究发现较高可靠性听觉图标警报系统在响应时间和主观可以进一步优化接管绩效和显示紧急程度，这是本项目研究的一个显著性特色成果。