Context-Awareness at the Service of Sensor Fusion Systems: Inverting the Usual Scheme

Abstract

Many works on context-aware systems make use of location, navigation or tracking services offered by an underlying sensor fusion module, as part of the relevant contextual information. The obtained knowledge is typically consumed only by the high level layers of the system, in spite that context itself represents a valuable source of information from which every part of the implemented system could take benefit. This paper closes the loop, analyzing how can context knowledge be applied to improve the accuracy, robustness and adaptability of sensor fusion processes. The whole theoretical analysis will be related with the indoor/outdoor navigation system implemented for a wheeled robotic platform. Some preliminary results are presented, where the context information provided by a map is integrated in the sensor fusion system.