Advancing Autonomous Vehicle Takeovers through Human-Centred Design

Autonomous vehicles (AVs) are poised to transform the transportation landscape, offering improved safety, efficiency, and convenience. However, achieving these benefits hinges on addressing the complexities of human-vehicle interaction, particularly during transitional phases where control shifts between the vehicle and the driver. This thesis investigates these critical dynamics through three interconnected studies. The first study provides a comprehensive review of Takeover Requests (TORs) in Level 3 AVs, emphasizing the importance of human-centred design in ensuring safe and efficient transitions. It introduces a framework that integrates environmental monitoring, driver state assessment, and ergonomic considerations to optimize handovers. The second study explores the influence of intersections on driver stress by analyzing heart rate (HR) data. Findings reveal that intersections could significantly heighten driver cognitive load, underscoring the need for AV systems and urban infrastructure to address these stressors. The third study leverages physiological signals, including electrocardiogram (ECG), electrodermal activity (EDA), and respiration (RSP), to model takeover quality using machine learning. The study demonstrates that these signals are robust predictors of driver readiness and performance during TORs, offering a pathway toward adaptive and safer AV systems. Together, these studies advance the understanding of human factors in autonomous driving, highlighting the critical role of physiological responses in designing intuitive, human-centric AV systems. This work contributes to the development of safer and more efficient mobility solutions, bridging the gap between technological innovation and human acceptance.