Many
sensor systems such as surveillance systems or target tracking systems can
be considered as stochastic dynamic processes involving multiple heterogeneous
components or agents, such as sensors, targets, and sensor platforms, and
environmental factors. A common formalism or mathematical representation
for these agents is necessary to automate and optimize any decision-making
process affecting system performance. Graphical models, such as decision
graphs, dynamic Bayesian networks ( DBNs
), and static Bayesian networks ( BNs ), can be
used to represent decision processes, sensor platforms and sensors
respectively, and can be unified in one mixed graphical model. Although
target and environmental characteristics are application dependent, a
general technique is being developed to integrate them in a BN sensor
model.
Conducting
the research at Duke University
is Dr.
Silvia Ferrari , Assistant Professor in Mechanical Engineering at Duke
University , and her graduate student, Chenghui Cai.