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

Capabilities

Reactivity

Finally, something these architectures can do. The basic philosophy of subsumption architectures is that they be very reactive to their environments. Since there is not knowledge representation, there is a short computational path between the sensor inputs and the output to the effectors. This results in highly reactive agents.

Interruptability

There are two ways to look at this issue of interruptability. We usually think of interrupts as interrupting a current task. Subsumption architectures aren't taskable, therefore it is easy to say they aren't interruptable.

However, subsumption architectures are a mass of constantly competing behaviors. Let's say an agent is currently performing a behavior such as "head for the object ahead" and suddenly someone places an obstacle in its path. At this point, the "avoid obstacles" behavior will become the new behavior. From this, we can see that Brooksian architectures are highly interruptable. In fact, they're completely interrupt-driven.

Navigation/Manipulation

Navigation is one of the basic behaviors found in all lower creatures, so naturally this capability would be found here. This was among the first behaviors Brooks endowed his agents with.

Coherent Behavior

This is rhetorical. The architectures are behaviorally-based approaches to intelligence, so they clear exhibit coherent behavior.

Perception