|
Abstract : |
It is well-known that planning is hard but it is less well-known how to approach the hard parts of a problem instance eectively. Using static domain analysis techniques we can identify and abstract certain combinatorial subproblems from a planning instance, and deploy specialised technology to solve these subproblems in a way that is integrated with the broader planning activities. We have developed a hybrid planning system (STAN4) which brings together alternative planning strategies and specialised algorithms and selects between them according to the structure of the planning domain. STAN4 participated successfully in the AIPS-2000 planning competition. We describe how sub-problem abstraction is done, with particular reference to route-planning abstraction, and present some of the competition data to demonstrate the potential power of the hybrid approach. 1, |
