基于智能驾驶场景的人机信任影响因素及作用机制研究  

The Mechanism of Influencing Factors of Human-Machine Trust in Intelligent Driving Scenes

作　　者：何文浩 王铁雁 张婷茹 陶达[1] HE Wen-hao;WANG Tie-yan;ZHANG Ting-ru;TAO Da(Institute of Human Factors and Ergonomics,Shenzhen University,Shenzhen 518060,China;Xiamen Meiya Pico Information Co.,Ltd,Xiamen 361008,China)

机构地区：[1]深圳大学人因工程研究所,广东深圳518060 [2]厦门市美亚柏科信息股份有限公司,厦门361008

出　　处：《人类工效学》2024年第2期9-16,共8页Chinese Journal of Ergonomics

基　　金：国家自然科学基金项目(32271130);广东省自然科学基金项目(24050000694,2023A1515012843);深圳市基础研究面上项目(JCYJ20230808105219038,JCYJ20210324100014040)。

摘　　要：目的研究智能驾驶场景下人机环系统中的三个重要因素(系统预测准确率、环境可认知状态和驾驶经验)对驾驶员智能驾驶辅助系统人机信任的影响。方法采用三因素组内组间混合设计的人因学实验,招募了24名不同驾驶经验的被试在智能驾驶辅助系统的三种预测准确率(90%、76%、60%)及两种环境可认知状态(清晰认知和模糊认知环境)下进行模拟驾驶任务,并收集驾驶过程中被试的主观人机信任、驾驶行为绩效和生理数据等多维指标,采用重复测量的方差分析处理实验数据。结果不同驾驶经验下,被试的主观信任存在显著差异。系统预测准确率和驾驶经验对主观信任存在交互作用。不同环境可认知状态下,被试的瞳孔直径、眼跳次数和皮电波峰幅值均存在显著差异。环境可认知状态和驾驶经验对眼跳次数存在交互作用。人机环三因素对被试的驾驶行为绩效无明显影响。结论系统预测准确率、环境可认知状态和驾驶经验均可在不同程度上影响主观人机信任及其相关生理指标。环境可认知状态较差时驾驶员会增加对驾驶辅助系统的信任。随着驾驶员驾驶经验的增加,其对驾驶辅助系统的信任也会增加。Objective The effects of three factors from the human-machine-environment system(i.e.,system prediction accuracy,environmental cognitive status,and driving experience)on drivers'trust towards an intelligent driving assistance system are examined.Methods A three-factor mixed design was employed in this human factors experiment.A total of 24 participants with different driving experience were recruited to perform simulated driving tasks under three accuracy levels of an intelligent driving assistance system(i.e.,90%,76%,and 60%)and two environmental cognitive states(i.e.,clear and ambiguous states).Multi-dimensional indicators such as subjective human-machine trust,driving behavior performance and physiological indicators during the driving tasks were collected.Repeated measures analysis of variance was used to examine the effects of the three independent factors on dependent variables.Results There were significant differences in subjective trust for participants with different driving experience.There was a significant interaction effect between system prediction accuracy and driving experience on subjective trust.There were significant differences in pupil diameter,number of saccades and skin conductance level in different environmental cognitive states.Environmental cognitive status and driving experience had significant interaction effects on mean number of saccades.The three factors had no significant impact on driving behavior performance.Conclusion The findings indicate that system prediction accuracy,environmental cognitive status,and driving experience can affect subjective human-machine trust and its related physiological indicators to varying degrees.It appears that,in poor environmental cognitive status or for more experienced drivers,drivers would trust more in the intelligent driving assistance system.

关 键 词：交通工程 人机共驾 智能驾驶 系统预测准确率 环境可认知状态 驾驶经验 人机信任 交通安全 

分 类 号：TP11[自动化与计算机技术—控制理论与控制工程]