Presentation
Effects of SA-Based Augmented Reality HUDs on Driver Situation Awareness in Level 3 ADS Under Adversarial Attack
DescriptionAn adversarial attack on Level 3 automated driving systems (ADS) introduces maliciously crafted perturbations to perception inputs, potentially leading to unsafe system decisions. To mitigate the safety risks posed by adversarial attacks and improve drivers’ situation awareness (SA) during takeover requests (ToR), this study proposes the implementation of SA-based augmented reality head-up displays (AR-HUDs). Forty participants took part in a laboratory experiment, completing a total of eight driving trials in a simulator under different SA-based AR-HUD conditions during adversarial attacks: baseline, SA1, SA2, SA3, SA1+SA2, SA1+SA3, SA2+SA3, and SA1+SA2+SA3. Driver SA was measured using the Situation Awareness Global Assessment Technique (SAGAT), and data were analyzed using linear mixed-effects models. The present study revealed a significant effect of AR-HUD conditions on drivers’ overall SA during L3 ADS takeovers under a road-surface induced adversarial attack. These findings provide experimental evidence that embedding SA level-specific cues within a multimodal ToR can enhance SA recovery under adversarial conditions. Furthermore, the results offer design implications for future multimodal ToR systems, highlighting the importance of incorporating SA level-specific cues to support more effective SA recovery.
Event Type
Lecture
TimeTuesday, October 20th2:10pm - 2:30pm PDT
Location