Information provided by a shared dynamic context model offers new possibilities in the realm of autonomous robots. The availability of external context information can be used by a robot to extend and to validate the locally acquired knowledge about its dynamic environment. A lot of helpful context information is already available in digital form and its quantity will increase rapidly, according to the vision of ubiquitous computing.

In this paper we present the Nexus Platform capable of managing a global dynamic context model that can be addressed and accessed in an easy and uniform way by all kinds of context-aware applications (like robots). Advantages for an autonomous wheelchair robot using the Nexus Platform are depicted within an airport scenario. We further present a framework for a sensor fusion agent, able to perform multi-sensor data fusion with selective attention control, concerning the current situation of the environment.