The report presents two standardized method logical approaches for evaluating the efficiency and safety of human-machine interaction in the use of partially automated driving functions. For this purpose, test criteria were defined to assess the fulfilment of the necessary requirements for conveying adequate system knowledge, adequate system and situation awareness, easy system operation and safe driver reaction at system boundaries. As a first method, an expert evaluation is used to check whether the human-machine interface (HMI, i.e. displays, system outputs and operating elements) meets the minimum requirements regarding the perceptibility and comprehensibility of system states. The focus is on visual and acoustic system outputs. In addition, aspects of conveying an appropriate understanding of the system via the user manual and system operation are evaluated. A checklist with a total of 14 requirements was developed to check these aspects. The procedure can be applied in a driving simulator, on test tracks or in real traffic. For this purpose, a set of use cases is experienced, on the basis of which the fulfilment of the individual items is checked. The checklist was tested and further developed in two real-life driving studies and one driving simulator study. The focus was on evaluating the validity (i.e. can the checklist distinguish between human-machine interaction concepts of different quality?) and the reliability (i.e. to what extent do different experts arrive at comparable results when evaluating a system?). As a second method, a standardized, scenario-based observation and survey tool for the evaluation of system interaction in the context of user studies was developed. As evaluation criteria, problems with system operation, driving behaviour and inadequacy of the monitoring behaviour are recorded, which lead to a superordinate examiner rating per use case. In addition, aspects of system understanding are recorded by questioning the driver about HMI-indicators and system reactions. These evaluation categories are implemented in a tablet application, the so-called S.A.D.E.-App (Standardized Application for Automated Driving Evaluation). The method was tested and validated in two simulator studies and under real traffic conditions with a series production vehicle. For a comprehensive, standardized evaluation of the human-machine interaction during partially automated driving, it is recommended to use both methods in combination. The procedure can be extended by additional methods to evaluate the long-term development of system trust and acceptance as well as behavioural adaptations. (Author/publisher)
Samenvatting