Philipp Kerschbaum

Designing take over scenarios for automated driving: How does augmented reality support the driver to get back into the loop?

Highly automated driving allows the driver to temporarily turn away from the driving task, meaning he or she does not have to monitor the system. This leads to the challenge of getting the driver back into the loop, if the automation reaches a system boundary. This study investigates, whether augmented reality information can positively influence the take over process. Therefore we evaluated two augmented reality concepts. The concept “AR red” displays a corridor on the road to be avoided by the driver in a take over scenario. The concept “AR green” suggests a corridor the driver can safely steer through. Results indicate that the type of augmented reality information does not influence take over times, but considerably affects reaction type. Visual inspection revealed higher consistency in driving trajectories for participants following the proposed corridor of “AR green” concept as compared to trajectories of drivers confronted with the restricted zone of “AR red”.

A transforming steering wheel for highly automated cars

In the near future, highly automated driving will almost certainly be available in commercial vehicles. Concerning the human-machine interface in such cars, two main issues have to be addressed. First, the detrimental effects of automation have to be avoided. Second, cars should provide an interface that allows the driver to utilize the time while driving highly automated. We conducted a driving simulator study to investigate the concept of geometrical transformation of the steering wheel to address both issues. Therefore, we implemented a prototype steering wheel which changes its shape depending on the current driving mode to improve mode awareness and comfort when driving highly automated. The study was focused on possible negative effects of the mechanical transformation in front of the driver during the take-over process. Results indicate that on average participants reacted faster and took over control later. The number of lane change errors, for example changing lanes without looking into the mirror, even somewhat decreased when using the transforming steering wheel. Furthermore, participants mainly rated the proposed concept as usable without problems during the take-over process.

Highly automated driving with a decoupled steering wheel

Future cars will almost certainly provide an increasing level of automation. Under certain conditions, they will allow the driver to withdraw from the control loop and deal with non-driving related tasks. To provide a convenient and safe user interface for this case, it can be advantageous to have the steering wheel de-coupled from the steering link and stationary. In this study, we evaluated two alternative steering wheel concepts. The first concept represents a state of the art steering wheel that decouples from the steering link and remains stationary at an angle of 0° during highly automated driving. In the second concept, the steering wheel shows the same behavior but does not have visible spokes. Hence, it does not display its physical orientation to the driver. Using a dynamic driving simulator, we evaluated the concepts in a comparison drive and a take-over scenario in a curve. A permanently coupled state of the art steering wheel served as control condition. Results show that the decoupling was only noticed by a small number of participants. Further, no negative impacts on the take-over process could be determined. The steering wheel with no visible spokes led to an even better performance compared to the control condition.

Designing the human-machine interface for highly automated cars — Challenges, exemplary concepts and studies

During the last years, intensive research has been conducted to make high degrees of automation available in cars. However, driver assistance systems today still need the driver to monitor the system. This will most probably change in near future, as highly automated driving becomes available. With the driver out of the control loop, this driving mode has beneficial aspects for the driver and could improve traffic safety as a high portion of traffic accidents are due to human error. On the other hand, high degrees of automation can have detrimental effects which are well known from other domains like aviation. These effects have led to fatal accidents in the past. The authors investigate various aspects of this topic and the corresponding challenges for the human-machine interface in future cars. In this paper, we present three of our research areas: the take-over process, trust in automation and utilization of drivetime. For each area, we explain theoretical background, current challenges and studies we conducted.