Meta-Reasoning Architecture Methodological Assumptions
Four key assumptions drove the design of the Meta-Reasoning Architecture
or MAX:
- Autonomous robots must mix
reaction with
planning;
- Limits on computational resources requires that
learning be directed by goals;
- Integration of capabilities should give priority to reaction, i.e.,
the agent must be situation-directed; and
-
Open,
homogeneous,
declarative knowledge,
separate control rules and
explicitly specified reasoning and learning operators can
facilitate the requisite capabilities.
Go to General Discussion of
Methodological Assumptions
Return to the top of this architecture.