School of Design, South China University of Technology, Guangzhou, China.
Research and Development Center of Transport Industry of Self-driving Technology, China Merchants Chongqing Communications Technology Research and Design Institute Co., Ltd., Chongqing, China; School of Big Data & Software Engineering, Chongqing University, Chongqing, China.
Accid Anal Prev. 2025 Jan;209:107826. doi: 10.1016/j.aap.2024.107826. Epub 2024 Nov 4.
Autonomous vehicles (AVs) should prioritise pedestrian safety in a traffic accident. External human-machine interfaces (eHMIs), which enhance communication through visual and auditory signals, become essential as AVs become prevalent. This study aimed to investigate the current state of research on eHMIs, with a specific focus on pedestrian interactions with eHMI-equipped AVs. A bibliometric analysis of 234 papers published between January 2014 and December 2023 was conducted using the Web of Science database. The analysis revealed a remarkable increase in eHMI research since 2018, with the principal research topics on crossing behaviour and eHMI evaluations of pedestrians. Subsequently, 38 articles were selected for a systematic review. The systematic review, conducted through a detailed examination of each selected article, showed that pedestrian crossing behaviour is usually measured using crossing initiation time, response time, walking speed and eye tracking data. The eHMI evaluations of pedestrians were made through questionnaires that measure clarity, preference and acceptance. Research findings showed that pedestrians' crossing behaviour and eHMI evaluations are influenced by human factors (age and nationality), vehicle factors (eHMI type, eHMI colour and eHMI position) and environmental factors (signalisation and distractions). The results also revealed that current eHMI experiments often use virtual reality and video methodologies, which do not fully replicate the complexities of real-world environments. Additionally, the exploration regarding the impact of human factors, such as gender and familiarity with AVs, on pedestrian crossing behaviour is lacking. Furthermore, the investigation of multimodal eHMI systems is limited. This review highlighted the importance of standardising eHMI design, and the key gaps in the current eHMI research were revealed. These insights will guide future research towards effective eHMI solutions through informed theoretical studies and practical applications in autonomous driving.
自动驾驶汽车 (AVs) 在交通事故中应优先考虑行人安全。外部人机界面 (eHMI) 通过视觉和听觉信号增强了通信，随着 AV 的普及，它变得至关重要。本研究旨在调查当前 eHMI 的研究现状，特别关注行人与配备 eHMI 的 AV 的交互作用。使用 Web of Science 数据库对 2014 年 1 月至 2023 年 12 月期间发表的 234 篇论文进行了文献计量分析。分析表明，自 2018 年以来，eHMI 研究显著增加，主要研究主题是交叉行为和行人对 eHMI 的评价。随后，对 38 篇文章进行了系统评价。通过对每篇选定文章的详细检查进行系统评价表明，行人的穿越行为通常使用穿越起始时间、反应时间、行走速度和眼动追踪数据来测量。行人对 eHMI 的评价是通过测量清晰度、偏好和接受度的问卷来进行的。研究结果表明，行人的穿越行为和 eHMI 的评价受到人为因素（年龄和国籍）、车辆因素（eHMI 类型、eHMI 颜色和 eHMI 位置）和环境因素（信号和干扰）的影响。研究结果还表明，当前的 eHMI 实验经常使用虚拟现实和视频方法，这些方法不能完全复制现实世界环境的复杂性。此外，缺乏对性别和对 AV 熟悉程度等人为因素对行人穿越行为影响的探索。此外，对多模式 eHMI 系统的研究也有限。本综述强调了标准化 eHMI 设计的重要性，并揭示了当前 eHMI 研究中的关键差距。这些见解将通过有针对性的理论研究和自动驾驶中的实际应用，指导未来的研究走向有效的 eHMI 解决方案。
