User and system within the context of use - how users cope with partial automation
Josten, Johanna; Ziefle, Martina (Thesis advisor); Eckstein, Lutz (Thesis advisor)
Aachen / RWTH Aachen University (2020, 2021) [Dissertation / PhD Thesis]
Page(s): 1 Online-Ressource : Illustrationen, Diagramme
In partial automation, the system takes over the active execution of the driving task in defined contexts of use, but relies on the driver to supervise both the environment and the system’s behaviour within the current driving situation. Currently, the driver has to supervise automation with hands on the steering wheel. While haptic feedback from the system has been considered beneficial for staying in the loop during automation use, hands-on supervision of automation has also been shown to reduce the comfort of use. This discrepancy between an assumed increase in the safety of use and a satisfying system design from a user perspective gave rise to investigations verifying the benefit of continuous hands-on supervision during automation use. A comprehensive analysis of users’ interaction with partial automation from initial contact to the repeated handling of control transitions was pursued in this work, with a focus on the role of haptic feedback. By variation of contexts of use, levels of user experience, abstraction levels of the driving task and in separate analyses for different interaction phases, the relevance of haptic feedback and the stability of its influence were assessed. In addition to performance after takeover requests, the investigation of post-automation driving performance constituted an exploratory focus in this work. Results from the a priori assessment of automation in a survey and automation use in four driving simulator studies were contrasted with a posteriori assessments and automation use in a test track study. Overall, hands-off monitoring was perceived as more comfortable. However, hands-on monitoring benefitted the supervision of automation and resulted predominantly in faster, more controlled transitions. The differences between the two feedback conditions were however small. Only monitoring quality, but not takeover timing differed between driving contexts. Post-automation driving performance reached manual performance levels shortly after control transitions. Differences between conditions were smaller on the test track and performance in transitions was more controlled than in the simulator studies. In general, the effects of automation in comparison to manual driving outweighed the influence of haptic feedback during automation use for subjective, gaze and performance metrics. For future studies, the voluntary choice for hands-off supervision and secondary task engagement under consideration of users’ mental models might provide further insights into the relevance of haptic feedback during use of partial automation.