A First-Order Formalization of Commitments and Goals for Planning - - PowerPoint PPT Presentation

a first order formalization of commitments and goals for
SMART_READER_LITE
LIVE PREVIEW

A First-Order Formalization of Commitments and Goals for Planning - - PowerPoint PPT Presentation

A First-Order Formalization of Commitments and Goals for Planning Felipe Meneguzzi 1 , Pankaj Telang 2 and Munindar Singh 2 1 Pontifical Catholic University of Rio Grande do Sul felipe.meneguzzi@pucrs.br 2 North Carolina State University


slide-1
SLIDE 1

A First-Order Formalization of Commitments and Goals for Planning

Felipe Meneguzzi1, Pankaj Telang2 and Munindar Singh2

1Pontifical Catholic University of Rio Grande do Sul

felipe.meneguzzi@pucrs.br

2North Carolina State University

Wednesday, 17 July 13

slide-2
SLIDE 2

Motivation

  • Commitments have been extensively studied in MAS
  • Encode high-level social relations between agents
  • Define communication protocols among agents
  • Previous formalizations
  • Operational semantics for goals and commitments,

and their interaction

  • Propositional planning formalization

Wednesday, 17 July 13

slide-3
SLIDE 3

Commitment Lifecycle

Expired (E) Null (N) Pending (P) Conditional (C) Detached (D) Terminated (T) Satisfied (S) Violated (V) Active (A) create antecedent failure antecedent cancel cancel ∨ consequent failure consequent release suspend reactivate

  • Formally

C(Debtor, Creditor, antecedent, consequent)

  • E.g.

C(buyer,seller,goods,paid)

Wednesday, 17 July 13

slide-4
SLIDE 4

Goal Lifecycle

Null (N) Inactive (I) Active (A) Suspended (U) Terminated (T) Failed (F) Satisfied (S) consider activate reconsider reactivate suspend suspend drop ∨ abort fail succeed

  • Formally

G(Agent, pg, s, f)

  • E.g.

G(buyer, needsgoods,goods,deadline )

Wednesday, 17 July 13

slide-5
SLIDE 5

Relating Commitments and Goals

  • Practical Rules relating commitments and goals
  • Let G = G(buyer,⊤,goods,⊥)

and C = C (buyer, seller, goods, pay)

  • Entice Rule: If G is active and C is null, buyer creates C
  • Motivation: Buyer can achieve its goals of goods by creating the

commitment to pay for them to Seller hGA, CNi create(C)

Wednesday, 17 July 13

slide-6
SLIDE 6

Hierarchical Task Network Planning

  • Generates a plan by successive refinement of tasks
  • Non-primitive Tasks - abstract, high-level tasks to be

decomposed

  • Primitive Tasks - cannot be further

decomposed (operators)

  • Multiple implementations

(e.g. JSHOP2, SHOP2)

  • Abstraction of choice for agent programming languages

nonprimitive task primitive task primitive task method instance

  • perator

instance

  • perator

instance precond precond precond effects effects s0 s1 s2

Wednesday, 17 July 13

slide-7
SLIDE 7

HTN Planning for Commitments and Goals

  • Formalization of commitment protocols in terms of HTN planning
  • Axioms enforcing state transition model

for goals and commitments

  • Planning Operators describing

transitions (e.g. create, suspend, etc.)

  • HTN Methods for practical rules

(e.g. entice, negotiate, etc.)

  • Allows HTN planner to be used to validate commitment protocols

HTN Planning Domain axioms methods

  • perators

Agent Goals Commitment Protocols HTN Planner Valid Enactments

Wednesday, 17 July 13

slide-8
SLIDE 8

A first-order formalization

  • Propositional formalization had several limitations
  • Limited expressivity
  • New First-order formalization:
  • Domain independent axioms, methods

and operators

  • Domain dependent

axioms, costs, methods and operators

  • Useful patterns of behavior

HTN Planning Domain domain axioms domain methods

  • perators

Agent Goals Commitment Protocols HTN Planner Multiple Enactments + Costs axioms methods domain

  • perators

Wednesday, 17 July 13

slide-9
SLIDE 9

Domain Independent Axioms & Operators

Commitment Axioms Goal Axioms

null(C, Ct, ~ Cv) ← ¬var(C, Ct, ~ Cv) conditional(C, Ct, ~ Cv) ← active(C, Ct, ~ Cv) ∧ ¬p(C, Ct, ~ Cv) detached(C, Ct, ~ Cv) ← active(C, Ct, ~ Cv) ∧ p(C, Ct, ~ Cv)

Commitment Operators

null(G, Gt, ~ Gv) ¬var(G, Gt, ~ Gv) inactiveG(G, Gt, ~ Gv) ¬null(G, Gt, ~ Gv) ^ ¬f(G, Gt, ~ Gv) ^ ¬s(G, Gt, ~ Gv) ^ ¬terminalG(G, Gt, ~ Gv) ^ ¬suspendedG(G, Gt, ~ Gv) ^ ¬activeG(G, Gt, ~ Gv) ~ ~

Goal Operators

hoperator !create(C, Ct, De, Cr, ~ Cv), pre(commitment(C, Ct, De, Cr) ^ null(C, Ct, ~ Cv)), del(), add(var(C, Ct, ~ Cv))i hoperator !suspend(C, Ct, De, Cr, ~ Cv), pre(commitment(C, Ct, De, Cr) ^ active(C, Ct, ~ Cv)), del(), add(pending(C, Ct, ~ Cv))i hoperator !consider(G, Gt, X, ~ Gv), pre(goal(G, Gt, X) ^ null(G, Gt, ~ Gv) ^ pg(G, Gt, ~ Gv)), del(), add(var(G, Gt, ~ Gv))i hoperator !activate(G, Gt, X, ~ Gv), pre(goal(G, Gt, X) ^ inactiveG(G, Gt, ~ Gv)), del(), add(activatedG(G, Gt, ~ Gv))i suspend ~

Wednesday, 17 July 13

slide-10
SLIDE 10
  • Axioms plus Domain-dependent operators
  • Commitment Axioms
  • Goal Axioms
  • Axioms Generated automatically using a compilation tool
  • Plus any domain-specific operators (e.g. purchase, ship, etc)

pg(G, Gt, ~ Gv) goal(G, Gt, X) ^ $ s(G, Gt, ~ Gv) goal(G, Gt, X) ^ & f (G, Gt, ~ Gv) goal(G, Gt, X) ^ # formula κ we define the rules below:

p(C, Ct, ~ Cv) ← commitment(C, Ct, De, Cr) ∧ ' q(C, Ct, ~ Cv) ← commitment(C, Ct, De, Cr) ∧ κ

en these two basic formulas from the commitment

Domain Dependent Definitions

Wednesday, 17 July 13

slide-11
SLIDE 11

achieveGoal(G1) create(C2) satisfy(C2) detach(C2) !paid(20, 123) pay(20, 123) !goods(123) create(C3) satisfy(C3) !paid(80, 123) pay(80, 123)

Patterns of Behavior

  • Concession Pattern

2 commitments

  • C2 - merchant commits to delivering the

goods upon a $20 payment from the customer

  • C3 - customer commits to pay $80 upon receiving the goods
  • By creating commitments C2 and C3, the customer has one possible

way of achieving its goal

Wednesday, 17 July 13

slide-12
SLIDE 12

Conclusions and Future Work

  • A FO formalization of goals and commitment protocols
  • Multiple interacting instances of the same goals and commitments
  • Piecemeal progress, concession, consolidation and compensation
  • Future Work
  • Reasoning about probabilities
  • Modelling non-cooperative partners

Wednesday, 17 July 13

slide-13
SLIDE 13

Questions?

Wednesday, 17 July 13