1 / 12

Logic Programs and Classical Planning with LTL Constraints

Logic Programs and Classical Planning with LTL Constraints. Ravi Palla Advisor:J Benton. Disjunctive Logic Programs (DLP). Consist of rules of the form A_1;A_2;…;A_k :- B_1,B_2,…,B_m,not B_{m+1},…,not B_n. A_i and B_i are atoms. Why LP encodings ?.

kylene
Télécharger la présentation

Logic Programs and Classical Planning with LTL Constraints

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Logic Programs and Classical Planning with LTL Constraints Ravi Palla Advisor:J Benton

  2. Disjunctive Logic Programs (DLP) • Consist of rules of the form A_1;A_2;…;A_k :- B_1,B_2,…,B_m,not B_{m+1},…,not B_n. A_i and B_i are atoms.

  3. Why LP encodings ? • Smaller Ground Instances compared to SAT encodings. • Answer Sets of DLP without negation are exactly the minimal models of the corresponding SAT encoding. This implies less redundant actions in the plans generated.

  4. Why LP encodings ? • Re-use of SAT encoding principles. • Linear time transformation to a logic program whose answer sets correspond to models – Answer set solvers just ground the program and use SAT solvers to compute answer sets.

  5. LP Encoding of the Domain • The complete initial state holds at level 0 and the end goals hold at the highest level. • Each fact can be transformed to a DLP rule by just re-writing all negative literals ~f as “ :- f. ”.

  6. LP Encoding of the Domain • Actions imply their preconditions and effects. a(T) -> precond(T) & effects(T+1) • Straight-forward translation to DLP rules. • Just Eliminate Conjunction in the consequent and move negative literals to the antecedent.

  7. LP Encoding of the Domain • Conflicting actions are mutually exclusive. ~a_1(T) v ~a_2(T) • Can be turned to :- a_1(T), a_2(T).

  8. LP Encoding of the Domain • Explanatory Frame Axioms f(T) & ~f(T+1) -> V_{a deletes f} a ~f(T) & f(T+1) -> V_{a adds f} a • Can be turned to DLP rules by just moving the negative literals to the consequent.

  9. LP Encoding of the PDDL 3.0 Constraints • Two different translations defined. • Naive but exponential. • Compact but uses new predicates.

  10. Compact Encoding for “always” • always F. • Turn F to DNF – F_1 v F_2 v … v F_k. • Add the following set of rules p(T) :- F_i(T) for all i. :- not p(T). • p is the new predicate introduced.

  11. Questions ?

  12. Thank You

More Related