Towards Interactive Driver Assistance System to Realize a Safe and an Efficient Driving

Abstract

Human interactions are related to the possibility to fail. Making errors or acting non optimal is strongly connected to the nature of the human being. The causes of errors resulting in preparation for and during interaction may be manifold. In case of formalizable interaction environments an action-discrete modeling approach can be used, distinguishing situations and operators describing formalizable scenes and related actions. The incomplete understanding of the situation is one of the reasons making errors. For successful interaction several options to get the final goal situation are often possible. As example, during driving, the current situation may confuse the driver to make the right decision. The driver may have the possibility to achieve the goal in optimal manner. Using an assistance system, the driver- vehicle interaction is studied based on the developed Situation-Operator-Modeling (SOM) approach. With the help of SOM, a situated action space is built to predict the intention of the human driver and to identify the optimal action sequence according to the given goal. The human driver can be warned when the detected intention includes dangerous elements or may lead to dangerous situations. The predicted next optimal action with corresponding goal can be displayed to improve the situation awareness and to realize a more reliable human-machine-interaction. An appropriately designed interface will close the loop between the human driver, the driving system, and the environment. After perceiving the displayed suggestion, the human driver may make decision whether to follow the instruction. The paper focuses on the establishment of the theoretical concepts introducing a qualitative and quantitative concept to integrate the vehicle, human driver, and interface as a loop to realize a safe and efficient drive.