Shaping Pedestrians’ Trust in Autonomous Vehicles: An Effect of Communication Style, Speed Information, and Adaptive Strategy

Abstract

AbstractCompared with conventional vehicles, autonomous vehicles (AVs) are featured by increased energy efficiency and road safety, yet hardly meet with much success without enough human trust. Designing appropriate interactions between AV and human, such as communication with pedestrians, could help enhance trust and public acceptance. In this work, we examine design characteristics of AV interface, including communication style, explicit display of vehicle speed, and adaptive strategy, and study their effects on pedestrians’ trust behaviors. It is found that any communication style could improve pedestrians’ trust in AV and decision alignment with AV expectations. Among the three communication styles, commanding and advisory are significantly better than informative, in terms of trust improvement (commanding versus informative: t = 3.61 and p < 0.001; advisory versus informative: t = 2.78, p = 0.005) and decision alignment ((a) in expected cross scenarios, commanding versus informative: t = 0.35 and p < 0.001; advisory versus informative: t = 11.71, p < 0.001; (b) in expected not cross scenarios, commanding versus informative: t = −7.61, p < 0.001; advisory versus informative: t = −6.40, p < 0.001). Adding speed information on top of explicit message communication does not change the relative effectiveness of individual styles, even though the display of speed-only information has significantly improved both measures (trust: F = 9.39 and p = 0.002; decision: F = 6.04 and p = 0.015). In addition, applying an adaptive communication strategy when yielding would significantly improve pedestrians’ trust (t = 9.33 and p < 0.001) and decision alignment (t = 14.78 and p < 0.001). This study demonstrates the influence of design characteristics on the formation of trust relationships between pedestrians and autonomous vehicles and paves the ways for developing more advanced AV communication mechanisms.