Highly Automated Driving in the Real World - A Wizard-of-Oz Study on User Experience and Behavior

摘要

Past research on human factors of automated driving has primarily focused on human performance aspects in takeover situations after partially or conditionally automated driving periods (SAE L2/3). In this study we investigated user experience and behavioral effects of driving in a highly automated (SAE L4) vehicle in mixed traffic on public roads without unplanned takeovers. The automation feature itself was realized by means of a Wizard-of-Oz approach, involving two trained operators for vehicle and HMI control. Twelve test participants were recruited to experience a series of four highly automated drives of about 30 minutes each. The main research questions were related to how driver attitudes towards automation and system trust develop over time as well as how this affects user behavior with respect to visual attention and engagement in non-driving related tasks (NDRT). The results show that the participants appreciated the automation feature for all road types (urban, rural and motorway scenarios). Positive attitudes towards automation started at an already high level and increased slightly over the course of the driving sessions. Manual analysis of the participants' individual behavior revealed large inter-individual differences for "total time eyes off road" and "frequency of glances towards driving related HMI". Participants did engage in NDRTs between 0 and 93% of the driving time, almost regardless of the driving environment. The explorative study also highlights the benefits and drawbacks of this particular research method. Although the Wizard-of-Oz illusion perfectly worked for all study participants, some critical issues remain to be solved for future applications.