EXCALIBUR
Adaptive Constraint-Based Agents in Artificial Environments

[REFINEMENT SEARCH]   [Total-Order Planning]   [Partial-Order Planning]   [Hierarchical Planning]   [Maximal Graphs]

[ Please note: The project has been discontinued as of May 31, 2005 and is superseded by the projects of the ii Labs. There won't be further updates to these pages. ]

Total-Order Planning

(Related publication: [PUBLink])

Early planning systems constructed plans in a so-called total order. Starting with an empty plan, in each refinement step there is a commitment to a new action at a specific plan position. This position must be in a total order with respect to the plan's other actions. The propagation method deduces resulting intermediate states and excludes the next refinement step's choice options for actions that do not contribute to satisfying unsatisfied preconditions of the plan's actions or goals. For a new action, choice options for plan positions are excluded if the position is after the action/goal with the corresponding unsatisfied precondition or if there would be another action between the new action and the action/goal with the corresponding unsatisfied precondition such that the required effect would be nullified. The search terminates successfully if all preconditions of the plan's actions and goals are satisfied. Progressive planning with forward chaining is also possible.

Examples of total-order planners are STRIPS [PUBLink], WARPLAN [PUBLink] and Waldinger's planner [PUBLink].


[REFINEMENT SEARCH]   [Total-Order Planning]   [Partial-Order Planning]   [Hierarchical Planning]   [Maximal Graphs]

For questions, comments or suggestions, please contact us.

Last update:
May 19, 2001 by Alexander Nareyek