Planning
- The system gets input from the user to set goals.
- Planning follows a neural implementation derived from a Maes net.
- Goals connect to modules.
- In connectionist terminology, these units have a continuous activation
value like IAT (sharks and jets).
- Modules also get input from facts (that are needed before they
fire), and can be inhibited or activated by other modules.
- If they are activated, the action is done.
- We do this with binary connectionist nodes (CAs) but are looking into
continuous ones.
- Ian Mitchell is now working on this, and has duplicated Maes'
original published system but in neurons.
