An Artificial Perceptual Model Approach - Artificial Human Sensors: Science and Applications - page 108

e.g., in a measuring chamber. Also possibleunbalancedsensorresponsetimewill
affect the measurement performance.
5.1.1 Traditional Sensor Fusion Models
Traditionally, a fusion process involves a hierarchical model that considers the sen-
sors and the sensor data as providing input values and delivering output results,
for example an identification. The details of how the process between the input
data and the resulting identification is formulated has in time varied among dif-
ferent proposed model approaches. The JDL-model, Hall (2001), is one of the most
well-known models that is based on a general approach to design its methodol-
ogy. The model was later improved by a fifth level user refinements, Blasch (2002).
Although this model proved advantageous to find a homogenous concept within
the field of sensor fusion, its shortcoming resides in the fact, that in many applica-
tions, more specificity are needed, e.g., use of extensive feedback and a dynamic
knowledge base, Biel (2002). Actually specific considerations may be taken when
designing a complex multi-sensor system:
The system requirement may involve parameters such as,
— accuracy achieved,
— resources needed,
—timerequired,
— flexibility of system, and
—costfactors.
The method or strategy involves sensor modeling and fusion methods in,
— measuring strategy,
— association of data,
— selection and extraction of features,
— validation of results, and
— of course, an extensive feedback function.
An active human related interaction can be introduced in a model and func-
tions as an effective feedback that recognises earlier experience implemented
phase of intelligence. By including the human function in an intelligent model,
there is an opportunity to enable an active perception model. In an artificial sensor
application, where the consideration is focussed on a symbiosis effect, the per-
ception process may contain both a human-like passive and active process. The
premises for passive systems, then the function lacks the ability to manage a mis-
sion and to act in the environment. On the other hand, by using an active sys-
tem there is a functionality that is able to momentarily change the initial mission
according to the occurrence in the environment, and thus redirect the sensors to
areas of interest or refine the common goal, Loutfi (2005).