Effect of Augmented Reality Cues in Autonomous Vehicles: A Comparative Study according to Environment Factors

Abstract

In the rapidly evolving landscape of autonomous vehicle technology, the integration of conditional autonomous vehicles (AVs) with human drivers poses distinct challenges, particularly in managing take-over requests (TORs) during driver engagement in non-driving-related tasks (NDRTs). This study investigated the effectiveness of augmented reality (AR) cues in enhancing TOR performance under varying conditions. Through a mixed-subject experiment with 42 participants, we employed a 2 (environments: day and night) × 4 (AR cue types: without, status, command, and combined cues) design framework. The research primarily concentrated on the differential impact of AR cue types (status vs. command) and their interplay with environmental factors on driving, cognitive, and NDRT performances. The results revealed a marked improvement in driving and cognitive performance with command AR cues, especially in night settings, where status AR cues were more effective. These findings underscore the critical role of environmental contexts in AR cue efficacy, suggesting the need for adaptive AR designs in AVs to optimize driver-vehicle interaction. This research not only contributes to the field of autonomous vehicle technology but also has broader implications for enhancing road safety and driver-vehicle collaboration in an increasingly automated future.