Quantitatively measuring user-experience with connected and autonomous vehicles in simulated virtual reality-based environments

The field of autonomous driving is sharply growing. One of the main issues related to such a context is the possible user’s lack of trust toward the autonomous vehicle. The area of Human-Machine Interaction (HMI) is expected to provide support in this context. Furthermore, physiological measures can help to obtain a real-time characterisation of the user’s physiological state, by letting us understand emotions like the stress level related to a specific situation and or task. Based on the above considerations, the aim of this thesis is to compare two different user interfaces which differ in the amount of information presented to the user: one is referred to “omni-comprehensive”, since it presents all the information which are expected to be available in connected and autonomous driving scenarios; the other one is named “selective”, as it exploits only a subset of that information. Physiologic measures, i.e. electrodermal activity (EDA) and heart rate variability (HRV), are used as indicators of affective states, and a driving experience in virtual reality is used to elicit changes in user’s stress levels. In particular, during the experience, seven unexpected events have been programmed to occur, representing different hazardous situations. A user study has been carried out, including also a pre-post questionnaire survey. The most substantial outcomes of the study show statistically significant differences in EDA between the two interfaces (p=.036) when analysing features extracted from EDA during the 10 seconds before and after the unexpected events and in several other control events. Unexpected events have a significant effect on EDA in both user interfaces (p<.001). Each user interface has shown up a strong relationship (p<.05 in six unexpected events) between the user’s subjective response and EDA. Data suggest that, even though the amount of the information provided by the omni-comprehensive interface has been considered as excessive compared to the selective one (p<.05), it can be regarded as contributing at reducing the stress response related to a pre-programmed hazardous event.