Jump to the Brooksian Overview page.
Jump to the Comparative Properties page.
Jump to the Home page.

Properties



Knowledge Representation

There are no explicit representations, symbolic or otherwise,  in a
subsumption architecture.  This is due to Brooks' philosophy
that intelligent behavior can be attained without explicit knowledge
representation.


A direct result of this is that there is no global world model.
In these architectures, it is believed that the world would be it's
own model.  This promotes the reactivity that is desired from these
agents, but it comes at the cost of not being able to plan.



Learning

From our reading, we gathered that learning isn't available in this
architecture as yet.  He has experimented some with simple learning,
but we haven't seen those results published.



Organization

As was stated above, this architecture is completely parallel and
distributed.  Nothing is centralized.  There is no central memory
store.  There is no global communications network.



General Performance

Agents based on this architecture are highly reactive, which is due to the
environments they're meant to work in.



Other