Context modeling for dynamic configuration of automotive functions

Current vehicles are usually equipped with an abundance of advanced driver assistant systems. Only a limited number of them can really be active permanently. The utility of the others depends on particular context scenarios. This motivates our goal of providing the car with the means necessary to dynamically adapt the set of active functions to its current requirements. Such a context-aware system has to construct a sound model of the actual context, based on available sources of information, such as sensors. In this paper, we present a generic context modeling approach suitable for dynamic configuration of automotive functions. The context model is divided into layers of different abstraction levels to enable the system to extract relevant context information. Data abstraction is accomplished by applying qualitative modeling techniques. The proposed method is sufficiently generic and enables an easy adjustment to specific system configurations and adaptation to new functions. The demonstration of the feasibility of the proposed solution and evaluation of its effectiveness was based on a simulated prototypical system configuration. Characteristics of ADAS functions were specified and their activation was measured during norm cycle test drives. The simulations yielded to a significant reduction in average function activity of an exemplary car system. Depending on the provided context parameters, a reduction of up to 24% was achieved.