Prediction of a system’s stress in succession to a human-machine interaction is difficult due to the variety and variability of the involved factors. Thereby, the human factor represents an important role, positive as well as negative, whereat the resulting uncertainty can be ascribed to the human performance variability. Current approaches for the investigation of the human influence onto system stress predominantly focus on human error and thus only on the negative aspects. In contrast, the concept of uncertainty recently attracts increased attention and allows for a holistic assessment of human induced uncertainty, but misses an applicable method. Assessment of the human influence onto the uncertainty during usage would lead to the reduction of safety measures and thus to a conservation of resources.

The present work addresses the development of a holistic approach for the characterization, assessment, quantification and control of the human influence onto the uncertainty during usage. Based on a literature review, a model for the description of human-machine interaction, focusing on human sub-processes, is developed and a total of 67 influencing factors are allocated to the model’s elements. On this basis, the method of Human Uncertainty Modes and Effects Analysis (HUMEAn) is derived, which allows for a systemic assessment and quantification of human induced uncertainty. The developed method of HUMEAn is subsequently applied within a laboratory study to investigate the uncertainty of the human sub-process execution of action. For this, 58 participants must fulfill the task to place a specific weight on top of a tripod. The interindividual human influence, represented by the strength and dexterity of the participants, as well as the influence of task variation in form of different placing weights and instructions, are assessed. As a first result, system stress seems to follow a lognormal distribution. Thereby, a significant negative influence of the placing weight as well as the strength of the participants onto the resulting system stress is found. In contrast, specific instructions as well as the dexterity of the participants show a significant positive impact onto uncertainty. During a second study with 44 participants, the HUMEAn is applied for the investigation of the complex task of landing an airplane. Thereby, the human sub-process choice of action in conjunction with intraindividual influences are focused. The uncertainty of choice of action is quantified by means of a Markov model. Again, the resulting uncertainty is represented by a lognormal distribution. Further, pilots holding a commercial pilot license tend to less variation within their action sequence as other pilots. Overall, a significant positive influence of the factors qualification, simulator- and flight experience are found. Moreover, several predictors for the resulting system stress for specific states of the Markov model are identified. A third study with 32 participants is conducted to investigate the applicability of appropriate interface design for the reduction of uncertainty. Therefore, participants must stack two identical weights consecutively on top of a tripod. The findings confirm the possibility to reduce uncertainty regarding the resulting system stress through the implementation of appropriate feedback.

Overall, the developed model and the derived methodological approach of the HUMEAn allow for a systematic and holistic characterization and quantification of human induced uncertainty. Based on the application of the method, implications for the control and reduction of human induced uncertainty can be realized, e.g. through selection or qualification of the operator as well as through appropriate interface design.