Knowledge Representation in Automatic Planning Systems
Stanford Research Institute Menlo Park United States
Pagination or Media Count:
This paper is a tutorial on automatic planning systems with particular emphasis given to knowledge representation issues. The effectiveness of a planner depends, to a large extent, on its ability to make use of descriptions and expertise associated with particular task domains. Examples of such domain knowledge include action models, state description models, scenarios, and special purpose plan composition methods. Our discussion focuses on how such knowledge is represented in planning systems and on how various planning strategies make use of it.