Information Technology Reference
In-Depth Information
«event»
E_s1
«sensor»
s_1
«effector»
e_1
«event»
Control
«event»
E_e1
«effector»
Control
«event»
Control
«event»
E_ek
«event»
E_sn
«sensor»
s_n
«effector»
e_k
Figure 6.10:
Mapping of a reactive agent architecture onto the GRAMS
reference model.
6.9 Evaluation
In this section the applicability of the GRAMS reference model is
evaluated. Firstly, it is demonstrated how basic agent architectures
are modeled in terms of the GRAMS reference model. Secondly, the
practical applicability of the GRAMS reference model is evaluated
using three case studies. Finally, a brief summary of the evaluation
results is given.
6.9.1 Mapping of basic agent architectures
The GRAMS reference model is intended as building block for more
sophisticated abstractions. In order to demonstrate this property of
the GRAMS reference model, three basic agent architectures (reactive,
deliberative, and reflective agent architectures) which represent a
broad range of agent architectures currently often used are mapped
onto the GRAMS reference model. An overview over a broad variety
of agent architectures may be found in [90].
Reactive agent architectures
Reactive agents are characterized by two aspects:
1. Reactive agents do not possess an internal model of the world.
2. The behavior is completely determined by the latest sensor input,
i. e., they react directly to their current sensor inputs.
