An Artificial Perceptual Model Approach - Artificial Human Sensors: Science and Applications - page 116

Figure 5.8. Description of the interactions in a generic sensor-based perception model with
a knowledge-based unit.
The sensor based perception model in Fig. 5.8 can also be seen from the per-
spective of interfacing between modules or sub-processes. The advantage of a
well-defined interface structure is of interest both in the sense of reusability and for
defining input/output criteria of each module or sub-process. The basic design is
aimed to view the human correlation and to strengthen the persuasive ideas about
effective perceptual sensor systems that are able to make powerful interactions.
In the interface between the model and the environment, a main objective is to
define the necessary sensors to obtain enough information from the area of interest.
Secondly, a request to focus on a certain object or event in the environment will
have the benefit to control and direct the sensors to an active area of events that
improve the ability to reach the goal of the mission, i.e., active perception.
In the procedure between the sub-activities sensation and perception, data is
arranged in a dimensional reduced organisation or in a filtered procedure and
placed in a structure, i.e., a data array, in order to handle both the separated data
as well as the overall time-tagged information package.
The interaction between the perception process, which is considered to be
passive per se, and its active counterpart, is a double directed communication
structure. The optional communication process directs the perceptual capability
in handling and organising the feature selection. This is suitable for the choos-
ing of measurement policy when deciding the structural perceptual procedure of
the sensor data collection in an environmental picture. The measurement policy is
considered to be of central activity in the model approach.
An important feature in the model is the knowledge-base process, which acts
as an overall involvement unit in the model. The interface to the knowledge base
is double directed, indicating that information gained from the process is contin-
uously updating knowledge of the system. Knowledge gained earlier can then
be obtained as additional information in future decision-making situations. This
quality will of course then increase the system performance.
In the other direction, the information provided to the different sub-
procedures is a powerful quality that affects the interaction between the different