Abstract Manager

Self-driving vehicles (SDVs) are promoted with the promise of improving traffic safety by eliminating crashes caused by human error, but there are concerns about how they may impact active travel. To reap the safety benefits of SDVs without discouraging active travel, responsible introduction of SDVs requires both maturity of SDV technology and acceptance by the public. This study addresses the latter part. 

We aim to gain a better understanding of how road users perceive interactions with SDVs, and recommend strategies for the responsible introduction of SDVs on public streets. Positing that pedestrian interactions with SDVs would be perceived differently than otherwise similar interactions with human-driven vehicles (HDVs), we call this differential “autonomy bias”. The objective of this study was to examine the relationship between personal attributes and autonomy bias. 

Web survey data were collected from participants throughout British Columbia, Canada in early 2022. Participants first answered questions about their socio-demographics, travel habits, and attitudes toward risk, general technology, and SDV technology. Participants then watched 8 short video clips of pedestrians interacting with motor vehicles in a crosswalk, and were prompted to rate the comfort and safety of the pedestrian and yielding of the motor vehicle. All 8 videos were actually HDVs that looked like SDVs (late-model dark-coloured sedans), but we deceived participants by randomly identifying 4 as showing “self-driving vehicles” in survey prompts. Based on ratings of comfort, safety, and yielding, we extracted three measures of autonomy bias for each participant using regression analysis. We then examined relationships between autonomy bias and personal characteristics with a structural equation model. 

Results indicate significant autonomy bias varying widely across participants, ranging from very positive to very negative impacts of self-driving technology on perceptions of comfort and safety. Among other factors, positive autonomy bias was associated with participants who reported being more familiar with SDV technology. Negative autonomy bias was associated with participants who reported being cis-men and having higher educational attainment, while being a visible minority had no significant effect.  Autonomy bias was more negative for people who walk or drive more frequently and positive for those who cycle more frequently.

The study results suggest that transportation agencies introducing SDVs should recognize the socio-demographic and other factors influencing how people perceive sharing the road with SDVs. Intervention strategies to improve autonomy bias should focus on familiarizing people with SDVs to alleviate anxiety about SDV technology (e.g., demonstrating safety performance of well-developed SDVs).