AAAI-20061 of 20 Deconstructing Planning as Satisfiability A Revised Report Henry Kautz UW University of Rochester in collaboration with Bart Selman and.

Slides:



Advertisements
Similar presentations
Constraint Satisfaction Problems
Advertisements

November 20GEOINFO 20061/25 A Robust Strategy for Handling Linear Features in Topologically Consistent Polyline Simplification Department of Computer Engineering.
Signal/Background Discrimination Harrison B. Prosper SAMSI, March Signal/Background Discrimination in Particle Physics Harrison B. Prosper Florida.
User meeting June Monthly and Seasonal forecasts Laura Ferranti and the Seasonal Forecast Section User meeting June 2006.
Fast optimal instruction scheduling for single-issue processors with arbitrary latencies Peter van Beek, University of Waterloo Kent Wilken, University.
REVIEW : Planning To make your thinking more concrete, use a real problem to ground your discussion. –Develop a plan for a person who is getting out of.
CSE391 – 2005 NLP 1 Planning The Planning problem Planning with State-space search.
Unit of EPIDEMIOLOGY SCIENTIFIC INSTITUTE OF PUBLIC HEALTH HEALTH EXPECTANCY INDICATORS IN BELGIUM 4th Meeting of the Task Force on Health Expectancies.
1 23 maart 2006 Surface construction techniques for volumetric objects How to maintain convex and concave features? Eddy Loke and Erik Jansen.
Where ya goin’, man? AOS Football Meeting August 14, 2006.
University of Toronto Mechanical & Industrial Engineering An Introduction to Constraint Programming J. Christopher Beck Dept. of Mechanical & Industrial.
Prepared by S. Saterfield
Experiments with RAINS-NL Jan Aben. Jan Aben, Experiments with RAINS-NL, Rome, May Contents Features and objectives of RAINS-NL Illustration.
Vermelding onderdeel organisatie 20/11/ DCA theme 3: Airport Systems Prof.dr.ir. Gabriel Lodewijks DCA - Delft Centre for Aviation Symposium Delft.
Magazine Journalism Welcome back to JO /10/2014template from copyright Overview Take attendance Review readings Write.
EAR-BASED AMENDMENT FORUM. September PROCESS AND PROCEDURES From Preparation of an Amendment to a Finding of “In Compliance”
EAR-BASED AMENDMENT FORUM. September Sponsored by the Pinellas Planning Council September 12 & 13, 2006 Harborview Center Clearwater.
THHGCS07B Coordinate Marketing Activities Lecture 2.
FAA AAR-410 December 5, FAA Airport Pavement Roughness R&D u Gordon Hayhoe, AAR-410, FAA William J. Hughes Technical Center, Atlantic City, New Jersey,
Data accreditation standard for the IM&T DES12 Sept The IM&T DES Using the tools that support e- audit John Williams & James Barrett.
CMPE 511, Fall Hybrid (Software-Hardware) Dynamic Memory Allocator Prepared by Mustafa Özgür Akduran İstanbul, 2006 Boğaziçi Üniversitesi.
More on Canonical Huffman coding. Gabriele Monfardini - Corso di Basi di Dati Multimediali a.a As we have seen canonical Huffman coding allows.
Planning Module THREE: Planning, Production Systems,Expert Systems, Uncertainty Dr M M Awais.
CLASSICAL PLANNING What is planning ?  Planning is an AI approach to control  It is deliberation about actions  Key ideas  We have a model of the.
1 Constraint Satisfaction Problems A Quick Overview (based on AIMA book slides)
State Transition Systems  linear planning  bounded model checking  conditional planning  model checking  state transition description languages: oPDDL.
1 Finite Constraint Domains. 2 u Constraint satisfaction problems (CSP) u A backtracking solver u Node and arc consistency u Bounds consistency u Generalized.
Dana Nau: Lecture slides for Automated Planning Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License:
1 Backdoor Sets in SAT Instances Ryan Williams Carnegie Mellon University Joint work in IJCAI03 with: Carla Gomes and Bart Selman Cornell University.
Ryan Kinworthy 2/26/20031 Chapter 7- Local Search part 1 Ryan Kinworthy CSCE Advanced Constraint Processing.
CSE 5731 Lecture 19 Reasoning about Time and Change CSE 573 Artificial Intelligence I Henry Kautz Fall 2001 Salvador Dali, Persistence of Memory.
CSE 5731 Lecture 21 State-Space Search vs. Constraint- Based Planning CSE 573 Artificial Intelligence I Henry Kautz Fall 2001.
1 Planning. R. Dearden 2007/8 Exam Format  4 questions You must do all questions There is choice within some of the questions  Learning Outcomes: 1.Explain.
1 Backdoors To Typical Case Complexity Ryan Williams Carnegie Mellon University Joint work with: Carla Gomes and Bart Selman Cornell University.
1 BLACKBOX: A New Paradigm for Planning Bart Selman Cornell University.
1 BLACKBOX: A New Approach to the Application of Theorem Proving to Problem Solving Bart Selman Cornell University Joint work with Henry Kautz AT&T Labs.
Reviving Integer Programming Approaches for AI Planning: A Branch-and-Cut Framework Thomas Vossen Leeds School of Business University of Colorado at Boulder.
Knowledge Representation II (Inference in Propositional Logic) CSE 473 Continued…
Ryan Kinworthy 2/26/20031 Chapter 7- Local Search part 2 Ryan Kinworthy CSCE Advanced Constraint Processing.
Sampling Combinatorial Space Using Biased Random Walks Jordan Erenrich, Wei Wei and Bart Selman Dept. of Computer Science Cornell University.
Planning as Satisfiability CS Outline 0. Overview of Planning 1. Modeling and Solving Planning Problems as SAT - SATPLAN 2. Improved Encodings using.
CS444-Autumn of 20 Planning as Satisfiability Henry Kautz University of Rochester in collaboration with Bart Selman and Jöerg Hoffmann.
Logical Foundations of AI Planning as Satisfiability Clause Learning Backdoors to Hardness Henry Kautz.
1 The LPSAT Engine and its Application to Metric Planning Steve Wolfman University of Washington CS&E Advisor: Dan Weld.
Heavy-Tailed Phenomena in Satisfiability and Constraint Satisfaction Problems by Carla P. Gomes, Bart Selman, Nuno Crato and henry Kautz Presented by Yunho.
Planning as Propositional Satisfiabililty Brian C. Williams Oct. 30 th, J/6.834J GSAT, Graphplan and WalkSAT Based on slides from Bart Selman.
Explorations in Artificial Intelligence Prof. Carla P. Gomes Module Logic Representations.
Dana Nau: Lecture slides for Automated Planning Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License:
CS 5411 Compilation Approaches to AI Planning 1 José Luis Ambite* Some slides are taken from presentations by Kautz and Selman. Please visit their.
Boolean Satisfiability Present and Future
Dana Nau: Lecture slides for Automated Planning Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License:
AAAI of 20 Deconstructing Planning as Satisfiability Henry Kautz University of Rochester in collaboration with Bart Selman and Jöerg Hoffmann.
Local Search Methods for SAT Geoffrey Levine March 11, 2004.
Inference in Propositional Logic (and Intro to SAT) CSE 473.
By J. Hoffmann and B. Nebel
Dana Nau: Lecture slides for Automated Planning Licensed under the Creative Commons Attribution-NonCommercial-ShareAlike License:
Inference in Propositional Logic (and Intro to SAT)
CS 4700: Foundations of Artificial Intelligence
Planning as Satisfiability
Planning as Search State Space Plan Space Algorihtm Progression
Local Search Strategies: From N-Queens to Walksat
Class #17 – Thursday, October 27
CSE P573 Introduction to Artificial Intelligence Class 1 Planning & Search INTRODUCE MYSELF Relative new faculty member, but 20 years in AI research.
Planning as Satisfiability with Blackbox
Graphplan/ SATPlan Chapter
Emergence of Intelligent Machines: Challenges and Opportunities
Class #19 – Monday, November 3
Graphplan/ SATPlan Chapter
Graphplan/ SATPlan Chapter
Presentation transcript:

AAAI of 20 Deconstructing Planning as Satisfiability A Revised Report Henry Kautz UW University of Rochester in collaboration with Bart Selman and Jöerg Hoffmann

AAAI of 20 AI Planning Two traditions of research in planning: –Planning as general inference (McCarthy 1969) Important task is modeling –Planning as human behavior (Newell & Simon 1972) Important task is to develop search strategies

AAAI of 20 Satplan Model planning as Boolean satisfiability –(Kautz & Selman 1992): Hard structured benchmarks for SAT solvers –Pushing the envelope: planning, propositional logic, and stochastic search (1996) Can outperform best current planning systems Satplan (satz)Graphplan (IPP) log.a5 sec31 min log.b7 sec13 min log.c9 sec> 4 hours

AAAI of 20 Satplan in 15 Seconds Time = bounded sequence of integers Translate planning operators to propositional schemas that assert:

AAAI of 20 Plan Graph Based Instantiation initial state: p action a: precondition: p effect: p action b: precondition: p effect: p q a0a1 p0p1p2 b1 m0m1 q2 ==

AAAI of 20 International Planning Competition IPC-1998: Satplan (blackbox) is competitive

AAAI of 20 International Planning Competition IPC-2000: Satplan did poorly Satplan

AAAI of 20 International Planning Competition IPC-2002: we stayed home. Jeb Bush

AAAI of 20 International Planning Competition IPC-2004: 1 st place, Optimal Planning –Best on 5 of 7 domains –2 nd best on remaining 2 domains PROLEMA / philosophers

AAAI of 20 International Planning Competition IPC-2006: Tied for 1 st place, Optimal Planning –Other winner, MAXPLAN, is a variant of Satplan! CPT2MIPS-BDDSATPLANMaxplanFDP Propositional Domains (1 st / 2 nd Places) 0 / 11 / 13 / 2 0 / 3 Temporal Domains (1 st / 2 nd Places) 2 / 0

AAAI of 20 What Changed? Small change in modeling –Modest improvement from 2004 to 2006 Significant change in SAT solvers!

AAAI of 20 What Changed? In 2004, competition introduced the optimal planning track –Optimal planning is a very different beast from non- optimal planning! –In many domains, it is almost trivial to find poor- quality solutions by backtrack-free search! E.g.: solutions to multi-airplane logistics planning problems found by heuristic state-space planners typically used only a single airplane! –See: Local Search Topology in Planning Benchmarks: A Theoretical Analysis (Hoffmann 2002)

AAAI of 20 Why Care About Optimal Planning? Real users want (near)-optimal plans! –Industrial applications: assembly planning, resource planning, logistics planning… –Difference between (near)-optimal and merely feasible solutions can be worth millions of dollars Alternative: fast domain-specific optimizing algorithms –Approximation algorithms for job shop scheduling –Blocks World Tamed: Ten Thousand Blocks in Under a Second (Slaney & Thiébaux 1995)

AAAI of 20 Domain-Independent Feasible Planning Considered Harmful Solution Quality? Speed? General optimizing planning algorithms BestModerate Domain-specific optimizing planning algorithms HighFast Domain-independent feasible planning ??

AAAI of 20 Objections Real-world planning cares about optimizing resources, not just make-span, and Satplan cannot handle numeric resources –We can extend Satplan to handle numeric constraints –One approach: use hybrid SAT/LP solver (Wolfman & Weld 1999) –Modeling as ordinary Boolean SAT is often surprisingly efficient! (Hoffmann, Kautz, Gomes, & Selman, under review)

AAAI of 20 Projecting Variable Domains initial state: r=5 action a: precondition: r>0 effect: r := r-1 Resource use represented as conditional effects a1 r=5 r=4 a0 r=4

AAAI of ICAPS Benchmarks

AAAI of 20 Large Numeric Domains Directly encode binary arithmetic action: a precondition: r k effect: r := r-k a1 r k r2 1 r3 1 r4 1 r1 2 r2 2 r3 2 r4 2

AAAI of 20 Objections If speed is crucial, you still must use feasible planners –For highly constrained planning problems, optimal planners can be faster than feasible planners!

AAAI of 20 Constrainedness: Run Time

AAAI of 20 Constrainedness: Percent Solved

AAAI of 20 Further Extensions to Satplan Probabilistic planning –Translation to stochastic satisfiability (Majercik & Littman 1998) –Alternative untested idea: Encode action failure as conditional resource consumption Can find solutions with specified probability of failure-free execution (Much) less general than full probabilistic planning (no fortuitous accidents), but useful in practice

AAAI of 20 Encoding Bounded Failure Free Probabilistic Planning plan failure free probability 0.90 action: a failure probability: 0.01 preconditions: p effects: q action: a precondition: p s log(0.89) effect: q s := s + log(0.99)

AAAI of 20 One More Objection! Satplan-like approaches cannot handle domains that are too large to fully instantiate –Solution: SAT solvers with lazy instantiation –Lazy Walksat (Singla & Domingos 2006) Nearly all instantiated propositions are false Nearly all instantiated clauses are true Modify Walksat to only keep false clauses and a list of true propositions in memory

AAAI of 20 Summary Satisfiability testing is a vital line of research in AI planning –Dramatic progress in SAT solvers –Recognition of distinct and important nature of optimizing planning versus feasible planning SATPLAN not restricted to STRIPS any more! –Numeric constraints –Probabilistic planning –Large domains