Event

Welcome to Linda Pipkorn's doctoral dissertation! She will defend her doctoral thesis "The driver response process in assisted and automated driving".

Linda Pipkorn is a PhD student in the group Crash Analysis and Prevention within Vehicle Safety division since 2018. She has a bachelor’s and master’s degree in Mechanical Engineering/Applied Mechanics from Chalmers University of Technology. Her research focus is on human collaboration with automated vehicles and is within the EU-project L3pilot. The aim of the research is to model how automated vehicles should behave in specific situations to meet human expectations and not trespass human comfort boundaries.

• Opponent: David Abbink (Professor, Department of Cognitive Robotics, Delft University of Technology, The Netherlands
• Supervisor/ Examiner: Marco Dozza (Professor, Chalmers University)

Summary

This PhD thesis assesses the safety of two types of vehicle automation systems: assisted driving and automated driving. Assisted driving systems are already present in cars on public roads today. These systems help drivers to accelerate, brake, and steer their cars—but are dependent on the drivers to make up for potential limitations, such as failing to detect a stationary vehicle on the road ahead. On the other hand, an automated driving system is assumed to reliably handle the complete driving task under certain conditions (e.g., highway driving). When the conditions no longer apply, the system will notify the driver to take over the complete responsibility of the driving task. This thesis acknowledges that the human role is different in assisted and automated driving than in manual driving and investigates this new driver role: specifically, the consequences it may have on traffic safety. The novelty of the thesis, which used data collected on a test track and public road, lies in the detailed study of the timing and quality of the process that the human goes through when responding to a system limitation in assisted driving or transitioning back to manual driving after a period of automated driving. In contrast, the literature to date mainly consists of studies conducted in driving simulators focusing on a driver’s single response time. Importantly, this thesis demonstrates that what is safe for assisted driving is not necessarily safe for automated driving. Drivers coming out of a period of automated driving could handle the same event that was challenging for several drivers in assisted driving. This thesis also found that, in the same event, a hands-on-wheel requirement may not actually hasten the driver response in assisted driving. Moreover, the thesis could not confirm the safety concerns, such as delayed response to events and even crashing, that had previously been attributed to assisted and automated driving compared to manual driving. Potential reasons for the differing findings are discussed: they may stem from the difference in test environments and system designs included in the various studies. Finally, the thesis also emphasizes the importance of including drivers' visual behavior alongside the time it takes to deactivate the system when assessing the safety of automated driving. Overall, this thesis presents data-driven research that contributes to making current assisted driving systems and future automated driving systems safe.