Not Always Good: Mitigating Pedestrians’ Less Careful Crossing Behavior by External Human-Machine Interfaces on Automated Vehicles

Abstract

An external human-machine interface (eHMI) is a novel tool that facilitates efficient interactions between pedestrians and automated vehicles (AVs). However, as pedestrians become accustomed to communicating with AV through eHMI, there is a risk of exhibiting less cautious behavior, which can increase the likelihood of accidents. This study aimed to examine the effects of methods to mitigate the side effects of communication with AV through an eHMI. A total of 132 participants participated in a Virtual Reality experiment to investigate the existence of negative effects and to explore mitigation methods. We designed five experimental conditions to manipulate the tone of text messages (Allocentric: “After you,” Egocentric: “I’ll stop”) and test the effect of mitigation methods, such as hiding the text messages on the eHMI when the AV is stopped (No eHMI, eHMI-A, eHMI-E, HeHMI-A, HeHMI-E). The results revealed that pedestrians were more likely to exhibit less careful behavior in the eHMI condition than in the No eHMI condition. The disappearance of text messages on the eHMI prompted pedestrian attention allocation toward traffic situations, leading to a reduction in accident risk. In particular, the mitigation method was most effective for safe crossing when pedestrians were continuously exposed to the eHMI, presenting an egocentric message. Our findings contribute to the design of text-based eHMIs for pedestrian decision-making when crossing and enhancing traffic safety.

Keywords:

• Pedestrian crossing
• traffic safety
• automated vehicle
• external Human-Machine interface
• virtual reality

Acknowledgements

This paper is based on results obtained from a project, JPNP18012, commissioned by the New Energy and Industrial Technology Development Organization (NEDO). Some of the preliminary results reported in this paper was published in the Proceedings of the International Ergonomics Association Meeting, 2021. The authors thank Ayami Takeuchi for her help in data collection.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data will be available upon reasonable request (owing to privacy/ethical restrictions).

Additional information

Notes on contributors

Jieun Lee

Jieun Lee is an assistant professor in the Department of Safety Engineering at Pukyong National University. Jieun received her PhD in Engineering from the University of Tsukuba in 2020. Her research interests include designing human-machine systems with the understanding of human factors in various domains.

Tatsuru Daimon

Tatsuru Daimon is a professor in the Faculty of Science and Technology at Keio University. Tatsuru earned his PhD in Engineering at Keio University in 1995. His research interests include driver assistance of intelligent transportation systems.