Abstract
Autonomous driving systems can relieve the driver of driving in a variety of either monotonous or demanding situations, for example in congested traffic. Since the introduction of such systems seems to be achievable due to progress in sensor technology and environmental modeling, research activities on such systems are strongly increasing. This paper discusses challenges on testing and evaluating these systems in regarding technology, human factors and the impact on traffic safety and efficiency.
The publication of research activities from different researchers and companies all over the world (e.g. Google, Continental) has raised the awareness of automated driving. Vehicle road automation has become a major research field on many different levels. The introduction of new functions raise the research question, whether the existing development and assessment methodologies still fulfill the requirements or whether the development and evaluation process of these functions needs to be adapted to new requirements and challenges of automated driving.
On basis of a state of the art analysis a development and evaluation process for automated driving functions is introduced, in which the system, the vehicle, the driver and the traffic environment are considered. The proposed evaluation process covers also the dependencies between impact, acceptance and controllability. For example impact assessment must consider long time periods for automated driving applications. This leads to the issue that today’s mainly scenario based approaches for safety impact assessment cannot be applied directly to automated driving applications.
Apart from the development process, the technical evaluation and impact assessment, user-related evaluation in terms of human factors needs to be considered with a different focus. Main research questions concern the control loop driver – vehicle – environment and the resulting performance, if the driver is taken out of this loop completely. The transition of the driving task between the driver and the system in situations in which the system reaches its functional limits has to be investigated with respect to various aspects ranging from controllability via traffic safety to driver acceptance.
The main result presented in this paper is a modular approach of a development and evaluation process for automated driving. This process will also describe appropriated methods for validation during the development as well as for the final evaluation considering an impact assessment. The basis for the process will be a literature review and evaluation of existing methodologies.
KEYWORDS – Driver interaction, automated driving, driver assistance