Jump to the RALPH-MEA Overview page.
Jump to the Comparative Properties page.
Jump to the Home page.

Properties


Knowledge Representation

As described in the
architectural details, RALPH-MEA uses various types of knowlege, all of them appear to be declarative. The different types of modules are used, and shared, by the EAs. Knowledge is stored symbollically, altough these is also a lot of probabilistic knowlege (i.e. it uses a lot of probabilities about what might happen, and what might be a consequence of a particular action, etc.) Altough it was not made very clear, it seems that each type of knowledge is stored in its own format.Each format optimized for storing that type of knowledge.


Learning

The architecture does not learn any new domain knowledge by itself. It can, however, compile knowledge from one of the types into another.


Organization

The architecture is organized into four Execution Architectures, and one metalevel which arbitrates between them. The MEAs run in parallel and give their results to the metalevel who decides what action is best to take. This is explaind in the
architectural details section.


General Performance

The RALPH-MEA architecture is supposed to be fast, as fast as possible. The metalevel arbitrates among the MEAs so that, if a quick response is needed it will deliver one as soon as possible. This assumes that the metalevel has enough knowledge to determine which situations require prompt attention. The sensors are always working and giving information to the EAs. These will either use this information or discard it, depending on whether they need it or not. It was not made clear how this is done.


Other