Effect of Explanatory Interfaces on Safety and Trust
Description
Control transitions in Level 3 automated driving rely heavily on human-machine interfaces (HMIs), yet empirical insights into how explanatory HMIs (X-HMIs) shape driver trust, perceived safety, and takeover performance across pre-task, during-task, and post-task phases remain scarce. This driving simulator study (n = 64) employed a 2 (HMI type: basic B-HMI vs. explanatory X-HMI) × 3 (task phase) mixed design to compare objective takeover metrics (time, lateral stability) and subjective evaluations across planned and unplanned takeover scenarios. In terms of objective performance, X-HMI did not significantly reduce takeover time for either planned or unplanned takeovers (p = 0.295 and p = 0.340, respectively) , but markedly improved takeover quality: y-axis velocity deviation decreased by 46% (p < 0.001) and steering angle deviation by 58% (p < 0.001), indicating enhanced lateral stability and smoother control. Subjectively, X-HMI significantly elevated trust and perceived safety across all phases (all p < 0.001) with large effect sizes (Cohen’s d > 1.4). These findings suggest that X-HMIs enhance Level 3 automation safety and user experience not by expediting driver responses, but potentially by calibrating appropriate trust and contributing to stable control transitions. Phase-specific transparency—aligning explanatory content with the cognitive demands of pre-, during-, and post-task interactions—emerges as a core HMI design principle, providing a validated framework for developing user-centered automated driving systems.
Files
Steps to reproduce
Build driving simulation experimental platform, carry out driving task tests for subjects, collect vehicle status, human factor perception and subjective evaluation data, clean and standardize raw data to form the final dataset.
Institutions
- Tongji UniversityShanghai, Shanghai
- Wuhan University of TechnologyHubei, Wuhan