Comparing different approaches to notify critical events in the operational driving domain of autonomous vehicles using virtual reality simulation

The advent of autonomous vehicles (AVs) brings forth a paramount challenge for the automotive industry, encompassing the task of enhancing the overall comfort, enjoyment, and safety of travel. Specifically, in the context of conditional automated driving, the driver has the ability to participate in non-driving related tasks (NDRTs) while the vehicle operates autonomously. However, there are certain situations in which the system may encounter challenges, prompting a take-over request (TOR). In such cases, the driver is required to regain control of the vehicle until the situation is resolved. The lack of trust people have in the technology stands as one of the primary factors that may hinder a widespread adoption of autonomous vehicles. A possible solution to this issue consists of the implementation of suitable interfaces, enabling seamless communication between the driver (the passenger, in perspective) and the vehicle. These interfaces serve as a vital conduit, providing comprehensive information regarding the vehicle's status, intentions, and perceptive understanding of the surrounding environment. However, improving trust in the system goes beyond merely providing information to the occupants. Comprehending how to effectively communicate with them, including what information to provide, how to deliver it, and when to do so, is of utmost importance. This approach is essential to prevent overwhelming the occupants, as an overload of information can lead to excessive cognitive load or unwarranted stress. This thesis focuses, as a first step, on prototyping an on-board interface for AVs using virtual reality (VR) based on the current state of the art in the field. The interface consists of different components: an instrument cluster (IC) display that provides general information about the vehicle, as well as a navigation map and a world in miniature showing the vehicle and its surrounding; a touchscreen display that presents the NDRT; an augmented reality (AR) windshield display (WSD) which integrates screen-fixed elements and world-registered AR overlays; an auditory interface that provides, using speech and abstract sounds, explanations and alerts about incoming events. A further step of the thesis focuses on how to effectively convey those critical events that, while still manageable by the autonomous system and not prompting a TOR, may cause a sense of stress and discomfort to the driver (e.g., a jaywalker). Finally, by conducting a VR driving simulation utilizing a motion simulator, the interface is subjected to an evaluation, considering the possible approaches to represent such critical events.