Intelligent environments are supposed to simplify the everyday life of their
users. To reach this target, a multitude of sensors is necessary to create a
model of the current scene. The complete processing of the resulting sensor data
stream can exceed the available processing capacities and inhibit the real-time
processing of the complete sensor data. A possible solution to this problem is a
fast pre-selection of potentially relevant sensor data and the restriction of
complex calculations on the pre-selected sensor data. That leaves the question
of what is potentially relevant.

The topic of this thesis is the construction of a multi-modal attention control
in an intelligent environment. Attention is the process of selectively
concentrating on one aspect of the environment while ignoring others. Especially
the human attention is intensily studied in psychology and cognitive
neuroscience. For this reason the principles of the human attention are surveyed
and transfered into the new application area.

The basis is a 3-dimensional, multi-modal saliency model, which is constructed
out of potentially relevant sensor data. This model is the base to transfer the
two human mechanisms of attention into the intelligent environment: the act of
mentally focusing on one of several possible sensory stimuli, the so-called
covert attention, is realized as a selection of the sensor which is assumed to
have the best perception of the scene; the act of directing sense organs towards
a stimulus source, the so-called overt attention, is realized as a multi-camera,
active-vision system which optimizes the visual scene perception and the
ì-dimensional saliency model.

The attention control was implemented and evaluated in the practiceoriented
intelligent environment FINCA at the Robotics Research Institute.