Planning in Soar
Planning in Soar
Soar's decision cycle provides a
natural foundation for planning tasks. In solving a problem, Soar
attempts to find a set of operators that lead from a given state to a
goal state. This string of operators can be viewed as a plan to move
the agent from the initial to goal state. When Soar reaches an impasse, in that it does not know the
next operator, it forms a sub-goal to choose the next operator.
Within this sub-goal it can perform a look-ahead search, simulating the effect of
different operators on the current state, in order to determine which
operator to apply next.
Return to the top of this architecture.
Go to a discussion of this capability
for multiple architectures.
Current Location: Soar-Capabilities-Planning