Integration of Human Factors-Related Knowledge into Decision Support Systems Applied to Assisted and Automated Operating Vehicles Using Examples for Inland Vessels

In human-machine systems, human behaviors are the main contributor to the safety of the overall system. Recognizing upcoming critical situations or knowing about critical actions in advance would enable the design of a new generation of human-machine systems that allow a smooth/fluid transition from assistance and intervention to direct guidance of the system. For most professionally operated complex systems such as power plants, aircrafts, or even ships, the workflow of human operation is highly regulated and can be considered in a formalized manner, which is helpful to serve as underlying system model. Future automation systems that allow to incorporate human assistance and monitoring are based on detailed sensor- and model-based situation awareness in the sense of knowing the consequences of possible alternative actions. With existing individualized knowledge about preferences and experiences from previous interactions as well as human error rates (e.g., from literature), a new quality of humanmachine systems can be generated that focuses on reliability and safety as goals.

As example for such a new system in this contribution a Situation-Operator-Modeling (SOM) approach is used to describe the captain-vessel-interaction and to illustrate the rule-based behavior as a graph-based-model. SOMbased action spaces consisting of possible captain's behaviors leading to a meaningful desired final situation are online analyzed and evaluated with respect to unsafe and unreliable actions components and or sequences, so from the manifold of possible sequences the best options can be defined and suggested in advance, critical and harmful ones can be denoted as critical by warnings etc. The reliability of the action sequences included in the action space are evaluated using a probabilistic risk assessment method called human error probabilities (HEP). The reliability analysis of the captain's actions in real time, newly introduced in the paper, enables safer driving behavior, reduction of accidents and dangerous situations. The manner novelty consists in the identification of dangerous situations and the intervention by appropriate warning and interaction strategies of the assistance system. Based on experimental examples, the paper evaluates the action components considering literature knowledge in addition to the underlying modeling.

Keywords: Automated and assisted operating vehicles, Decision support system, Human-machine-interaction, Situation-operator-modeling, Human error probability, Action space.

Download PDF