JTF ATD Core Plan Representation Workshop

Austin Tate (a.tate@ed.ac.uk), 20-Jun-96
This page provides notes of Austin Tate's input for the July 1996 Workshop on the JTF ATD Common Plan Representation ( http://projects.teknowledge.com/CPR/cpr.html) which in turn may give input to the C2 Schema used in the JTF ATD.

A related information page is available here.

A description of my proposed approach to Mixed Initiative Planning (mutually constraining the space of behaviour) and its relationship to the <I-N-OVA> constraint model of activity is available here.

  1. If different applications have different views of planning, is it possible for them to externalize these views, so that they can operate on a common model? What are these different views, and what is this common model?

    Yes, so long as an appropriate hierarchical structure is employed. I suggest that we follow the Process Interchange Format (PIF) model of a small core representation of Process and Activity (largely compatible with IDEF) with its "Partial Shared View" (PSV) extension mechanism.

  2. How should a core plan representation support mixed initiative planning and partial plans?

    Adopt a model of plans being constraints on the legitimate behaviours in the domain. Mixed initiative is possible if all agents (system and human) involved in the planning process share a view of the planning process of mutually constraining the behaviours desired (by setting, analysing, modifying and using the constraints). This means having a model of activities that are possible in the domain, and an extendible representation for all the constraints that are possible on those activities.

  3. How should a core plan representation support uncertainty? Specifically, how can a user encode information about alternative actions, assumptions, predictability, and range of possible outcomes?

    Constraints naturally allow spaces of alternative elaborations of the partially specified plans.

  4. How should a core plan representation address the goals of both simplicity and scalability? How can the representation and user views support a range of planning needs?

    A core process/actrivity model extendible with partially shared views (shared between those systems needing the additional information, but sharing the core or deeper models as necessary). Separation of the model from model views is essential anyway, and can support technical plan views or world oriented simulation or animation views.

  5. What are the kinds of user interactions with plans that we should be thinking about? What does an individual need to do? What kinds of group dynamics need to be supported?

    We should be seeking an uniform interlingua between system components and between the various roles of users involved which seeks to communicate constraints on the mission taskings, options being explored, authority, world state, geographical/spatial, resource and other constraints. Ways to present this interlingua information in an acceptable form to the various agents (humans and systems) involved should be explored.

  6. What lessons can be learned from previous efforts at designing common plan representations?

  7. Are there existing tools and methods outside of JTF which the core plan representation should be compatible with?