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

Inabilities

Planning

Planning in current subsumption architectures is not possible since there is not explicit representation of any kind.

Prediction

Prediction, like planning, is a high-level activity that requires explicit knowledge representation, so it is not available in current subsumption architectures.

Meta-Reasoning

Meta-reasoning and meta-knowledge requires that knowledge be represented declaratively. Current subsumption architectures do not represent their knowledge at all, so it is not available in current subsumption architectures.

Taskability

To quote Brooks, "This is an orthogonal issue". This architecture is a behaviorally-based system. Unlike functionally-based system, these agents aren't designed to be given instructions. They are, rather, a collection of competing behaviors that drive the agent.

Learning

In current subsumption architectures, learning is not implemented. Like many of the capabilites listed above, some kind of knowledge representation is typically needed for a system to be able to learn.

NLP

Don't even think of it! No explicit symbolic representation means no NLP.