Designing Human-Machine Interaction for Trustworthy Automated Vehicles

Abstract

Automated vehicles are anticipated to not only improve safety and comfort but also redefine the nature of driver-vehicle interactions. With these advancements, drivers in automated vehicles share the monitoring and supervising role with the system, creating a unique human-machine team. In this relation, beyond technical excellence, it is important for drivers to comprehend and trust the system for its effective and safe utilisation. Trust evolves during interactions, influenced by the system’s performance, emphasising the significance of provided information and interaction format. Design features of driving automation, particularly user interfaces, impact trust and shape perceptions of system performance, forming a crucial element for human-machine interaction.

This dissertation addresses three critical research gaps. First, an understanding of driver-vehicle interaction is crucial for designing effective interfaces. Current studies lack a clear comprehension of interactions during driving and transitions in complex systems, emphasising the need for an integrated view that considers various contexts and levels of driving automation. In this dissertation, I address this gap by investigating how drivers perceive complex interactions and identifying necessary interactions for drivers. Second, interactions extend beyond isolated take-over events, forming a sequence of interconnected behaviours that shape the overall driving experience. While take-over situations are undeniably critical, they represent just one facet of a broader continuum of scenarios. In-depth research on interactions during automated driving, where the human driver transitions between passive monitoring and active engagement, is lacking. Therefore, I explore the interactions in various situations, including automated driving and mode transitions in automated vehicles. Third, my research emphasises the design phase of user interfaces, beyond the conventional focus on identifying the effects of interfaces on drivers in specific scenarios. It stresses the importance of considering interface design comprehensively, incorporating factors such as cognitive load, situational awareness, and driver experience. Specifically, I look into the design of soundscapes to guide the driver back to control in take-over situations, creating a novel transition experience while prioritising safety.

Aligned with the goal of designing human-machine interaction for trustworthy automated vehicles, the research objectives are delineated into two objectives and six studies. The first objective aims to understand the driver-automated vehicle interaction, focusing on identifying the effects of interfaces, investigating human-machine interaction, and understanding drivers’ mode transition logic. The second objective aims to contribute to the development of interaction design guidelines, designing and evaluating user interfaces and developing an approach to soundscape design in automated vehicles. Finally, I consolidate the findings, discuss the research’s contribution, and provide an outlook on future work in the evolving landscape of automated vehicle technology.

Six studies employing diverse methods yield following results. A literature review and an on-road study unveil that the interaction between the driver and the automated vehicle through user interfaces significantly influences drivers’ performance and trust. Specifically, the literature review indicates that this interaction during automated driving has an impact extending from take-over situations to overall performance. The on-road study, using a vehicle with multiple levels of driving automation, reveals mode confusion related to mode transitions. Further exploration of driver’s understanding of mode transition logic, through an online survey demonstrates that, despite the driver performing certain interventions related to driving functions, there is no dominant mental representation of mode-transition logic under specific scenarios. Simulator experiments in partial and conditional automated driving scenarios illustrate that providing automation information during driving enhances the driver’s trust and acceptance. In particular, delivering the manoeuvre of driving automation to the driver through auditory modality is effective in enhancing trust and acceptance. The proposed soundscape design for take-over requests and spatial sound delivering manoeuvre information while automated driving showcases the potential of sound design to elevate the driver’s experience beyond a simple beep.

Throughout this dissertation, I investigate interactions in automated vehicles, addressing interaction challenges and designing and evaluating user interfaces. By conducting a literature review, on-road observations, interviews, online surveys, and simulator experiments, it navigates the complexities of trust, acceptance, and overall user experience. The contributions extend to understanding driver trust, performance, and the impact of system complexities on human-machine interaction, enriching the field with empirical evidence and practical guidelines for interaction design.