Similarity may be safer: the effect of similarity between Speech-based takeover request style and driver personality on self-driving takeover performance

Abstract

In Level 3 automated driving, it is critical that drivers can rapidly and effectively shift from non-driving related tasks (NDRT) back to the driving task. While previous research has examined the modality, timing, and vocal characteristics of takeover requests (TORs), little is known about how the style of speech-based TORs interacts with drivers’ personality traits. This study conducted a driving simulator experiment with 49 participants using a 2 × 2 within-subjects design. Drawing on the dominant-submissive dimension of personality, we examined the similarity of personality tendencies between speech-based TORs and drivers under takeover scenarios of varying urgency (low: road construction; high: traffic accident). The findings reveal that personality similarity had a significant impact on both response time and takeover performance. When the speech-based TORs aligned with the driver’s personality tendencies, participants demonstrated shorter attention redirection times and more stable takeover quality. And high urgency scenarios elicited faster takeover responses but at the cost of reduced stability. Importantly, similar speech-based TORs were shown to mitigate this deterioration in takeover stability.  Future applications may consider integrating personality-matching Speech-based TORs into automated driving systems (ADSs). This study provides insights for designing human-machine interfaces (HMIs) for automated driving, contributing to improved takeover safety.