Research

In order to provide broad coverage of many different planning approaches, we represent a plan in terms of contexts, which capture planning background information that is independent of a particular planning episode, external constraints, for example user advice and preferences, and commitments made during the planning process.

The overview of the PLANET ontology below shows the central concepts and relations. Some arrows do not have targets, indicating relations whose ranges are not fixed by the ontology.

We have used Planet to represent plans in three real-world domains with very different features:

• In an air campaign planning domain, where plans are expanded by humans an a task-decomposition structure.
• In a COA analysis domain, where plans are entered by experts and tasks have explicitly entered purposes and optionally supporting tasks.
• In a workarounds domain to replace or repair damaged bridges, where plans were automatically generated and then evaluated by a separate tool.