Planning Graph-based Heuristics for Cost-sensitive Temporal Planning Minh B. Do & Subbarao Kambhampati CSE Department, Arizona State University

Slides:



Advertisements
Similar presentations
Suleyman Cetintas 1, Monica Rogati 2, Luo Si 1, Yi Fang 1 Identifying Similar People in Professional Social Networks with Discriminative Probabilistic.
Advertisements

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.
An LP-Based Heuristic for Optimal Planning Menkes van den Briel Department of Industrial Engineering Arizona State University
CS6800 Advanced Theory of Computation
Top 5 Worst Times For A Conference Talk 1.Last Day 2.Last Session of Last Day 3.Last Talk of Last Session of Last Day 4.Last Talk of Last Session of Last.
Finding Search Heuristics Henry Kautz. if State[node] is not in closed OR g[node] < g[LookUp(State[node],closed)] then A* Graph Search for Any Admissible.
Effective Approaches for Partial Satisfaction (Over-subscription) Planning Romeo Sanchez * Menkes van den Briel ** Subbarao Kambhampati * * Department.
Graph-based Planning Brian C. Williams Sept. 25 th & 30 th, J/6.834J.
Enhancing Search for Satisficing Temporal Planning with Objective-driven Decisions J. Benton Patrick EyerichSubbarao Kambhampati.
Planning Graphs * Based on slides by Alan Fern, Berthe Choueiry and Sungwook Yoon.
Lecture 10: Integer Programming & Branch-and-Bound
Dynamic Pickup and Delivery with Transfers* P. Bouros 1, D. Sacharidis 2, T. Dalamagas 2, T. Sellis 1,2 1 NTUA, 2 IMIS – RC “Athena” * To appear in SSTD’11.
Planning and Scheduling. 2 USC INFORMATION SCIENCES INSTITUTE Some background Many planning problems have a time-dependent component –  actions happen.
3/4  The slides on quotienting were added after the class to reflect the white-board discussion in the class.
Happy Spring Break!. Integrating Planning & Scheduling Subbarao Kambhampati Scheduling: The State of the Art.
3/25: Leaving STRIPS Planning and going to Sapa. Administrivia 3/25  Homework 4 due next class  Midterm soon after that  Will be take home  Will have.
AI – Week AI Planning – Plan Generation Algorithms: GraphPlan Lee McCluskey, room 2/09
Energy Management and Adaptive Behavior Tarek Abdelzaher.
Over-subscription Planning with Numeric Goals J. Benton Computer Sci. & Eng. Dept. Arizona State University Tempe, AZ Minh Do Palo Alto Research Center.
Planning: Part 3 Planning Graphs COMP151 April 4, 2007.
Extending Graphplan to handle Resources Presenter: Pham Van Cuong Department of Computer Science New Mexico State University.
TADA Transition Aligned Domain Analysis T J. Benton and Kartik Talamadupula and Subbarao Kambhampati.
Ryan Kinworthy 2/26/20031 Chapter 7- Local Search part 1 Ryan Kinworthy CSCE Advanced Constraint Processing.
4/8: Cost Propagation & Partialization Today’s lesson: Beware of solicitous suggestions from juvenile cosmetologists Exhibit A: Abe Lincoln Exhibit B:
10/18: Temporal Planning (Contd) 10/25: Rao out of town; midterm Today:  Temporal Planning with progression/regression/Plan-space  Heuristics for temporal.
1. 2 Problem Description & Assumption Metric model in Sapa planner:  f(p) = w * time(p) + (1-w) * cost(p).  Assuming that the trade-off value w is given.
Concurrent Probabilistic Temporal Planning (CPTP) Mausam Joint work with Daniel S. Weld University of Washington Seattle.
1001 Ways to Skin a Planning Graph for Heuristic Fun and Profit Subbarao Kambhampati Arizona State University (With tons of.
A Hybrid Linear Programming and Relaxed Plan Heuristic for Partial Satisfaction Planning Problems J. Benton Menkes van den BrielSubbarao Kambhampati Arizona.
Expressive and Efficient Frameworks for Partial Satisfaction Planning Subbarao Kambhampati Arizona State University (Proposal submitted for consideration.
Minh Do - PARC Planning with Goal Utility Dependencies J. Benton Department of Computer Science Arizona State University Tempe, AZ Subbarao.
Reviving Integer Programming Approaches for AI Planning: A Branch-and-Cut Framework Thomas Vossen Leeds School of Business University of Colorado at Boulder.
Integrating Planning & Scheduling Subbarao Kambhampati Integrating Planning & Scheduling Agenda:  Questions on Scheduling?  Discussion on Smith’s paper?
CS121 Heuristic Search Planning CSPs Adversarial Search Probabilistic Reasoning Probabilistic Belief Learning.
SAPA: A Domain-independent Heuristic Temporal Planner Minh B. Do & Subbarao Kambhampati Arizona State University.
Local Search Techniques for Temporal Planning in LPG Paper by Gerevini, Serina, Saetti, Spinoni Presented by Alex.
1 Combinatorial Problems in Cooperative Control: Complexity and Scalability Carla Gomes and Bart Selman Cornell University Muri Meeting March 2002.
Planning II CSE 573. © Daniel S. Weld 2 Logistics Reading for Wed Ch 18 thru 18.3 Office Hours No Office Hour Today.
Classical Planning Chapter 10.
Embedded System Design Framework for Minimizing Code Size and Guaranteeing Real-Time Requirements Insik Shin, Insup Lee, & Sang Lyul Min CIS, Penn, USACSE,
CHALLENGING SCHEDULING PROBLEM IN THE FIELD OF SYSTEM DESIGN Alessio Guerri Michele Lombardi * Michela Milano DEIS, University of Bologna.
Homework 1 ( Written Portion )  Max : 75  Min : 38  Avg : 57.6  Median : 58 (77%)
STDM - Linear Programming 1 By Isuru Manawadu B.Sc in Accounting Sp. (USJP), ACA, AFM
© J. Christopher Beck Lecture 5: Project Planning 2.
ANTs PI Meeting, Nov. 29, 2000W. Zhang, Washington University1 Flexible Methods for Multi-agent distributed resource Allocation by Exploiting Phase Transitions.
USING PREFERENCE CONSTRAINTS TO SOLVE MULTI-CRITERIA DECISION MAKING PROBLEMS Tanja Magoč, Martine Ceberio, and François Modave Computer Science Department,
Probabilistic Reasoning for Robust Plan Execution Steve Schaffer, Brad Clement, Steve Chien Artificial Intelligence.
Efficient Provisioning of Service Level Agreements for Service Oriented Applications Valeria Cardellini, Emiliano Casalicchio, Vincenzo Grassi, Francesco.
Cost-Optimal Symbolic Pattern Database Planning with State Trajectory and Preference Constraints Stefan Edelkamp University of Dortmund.
Cost-Optimal Planning with Constraints and Preferences in Large State Spaces Stefan Edelkamp, Shahid Jabbar, Mohammed Nazih University of Dortmund.
1 Iterative Integer Programming Formulation for Robust Resource Allocation in Dynamic Real-Time Systems Sethavidh Gertphol and Viktor K. Prasanna University.
August 30, 2004STDBM 2004 at Toronto Extracting Mobility Statistics from Indexed Spatio-Temporal Datasets Yoshiharu Ishikawa Yuichi Tsukamoto Hiroyuki.
CONSTRAINT-BASED SCHEDULING AND PLANNING Speaker: Olufikayo Adetunji CSCE 921 4/08/2013Olufikayo Adetunji 1 Authors: Philippe Baptiste, Philippe Laborie,
A local search algorithm with repair procedure for the Roadef 2010 challenge Lauri Ahlroth, André Schumacher, Henri Tokola
© Daniel S. Weld 1 Logistics Travel Wed class led by Mausam Week’s reading R&N ch17 Project meetings.
Graphplan CSE 574 April 4, 2003 Dan Weld. Schedule BASICS Intro Graphplan SATplan State-space Refinement SPEEDUP EBL & DDB Heuristic Gen TEMPORAL Partial-O.
Search Control.. Planning is really really hard –Theoretically, practically But people seem ok at it What to do…. –Abstraction –Find “easy” classes of.
By J. Hoffmann and B. Nebel
Heuristic Search Planners. 2 USC INFORMATION SCIENCES INSTITUTE Planning as heuristic search Use standard search techniques, e.g. A*, best-first, hill-climbing.
EBL & DDB for Graphplan (P lanning Graph as Dynamic CSP: Exploiting EBL&DDB and other CSP Techniques in Graphplan) Subbarao Kambhampati Arizona State University.
Artificial Intelligence Chapter 11 Alternative Search Formulations and Applications.
Chap 4: Searching Techniques Artificial Intelligence Dr.Hassan Al-Tarawneh.
Dynamic Pickup and Delivery with Transfers
Debugging Constraint Models with Metamodels and Metaknowledge
Basic Project Scheduling
Title: Suggestion Strategies for Constraint- Based Matchmaker Agents
Chap 4: Searching Techniques Artificial Intelligence Dr.Hassan Al-Tarawneh.
Graph-based Planning Slides based on material from: Prof. Maria Fox
Chap 4: Searching Techniques
Heuristic Planning with Time and Resources
Presentation transcript:

Planning Graph-based Heuristics for Cost-sensitive Temporal Planning Minh B. Do & Subbarao Kambhampati CSE Department, Arizona State University

Motivation Multi-dimensional nature of plan quality in metric temporal planning: –Temporal quality (e.g. makespan, slack) –Plan cost (e.g. cumulative action cost, resource consumption) Necessitates multi-objective optimization: –Modeling objective functions –Tracking different quality metrics and heuristic estimation  Challenge: There may be inter-dependent relations between different quality metric

Example Option 1: Tempe  Phoenix (Bus)  Los Angeles (Airplane) –Less time: 3 hours; More expensive: $200 Option 2: Tempe  Los Angeles (Car) –More time: 12 hours; Less expensive: $50 Given a deadline constraint (6 hours)  Only option 1 is viable Given a money constraint ($100)  Only option 2 is viable Tempe Phoenix Los Angeles

General Problem Planner Good quality solution Problem specification Objective function How to design objective function? -User define -Learning users utility model We do not investigate We investigate Given the objective function that involve both time and cost quality  Finding heuristics that sensitive to the cost function

Our approach Using the Temporal Planning Graph (Smith & Weld) structure to track the time-sensitive cost function: –Estimation of the earliest time (makespan) to achieve all goals. –Estimation of the lowest cost to achieve goals –Estimation of the cost to achieve goals given the specific makespan value.  Using those information to calculate the heuristic value for the objective function involving both time and cost

Outline Action representation and Temporal Planning Graph Time sensitive cost functions: –Cost propagation using the temporal planning graph. –Termination criteria for the cost propagation process. Deriving heuristic values from cost functions –Direct calculation –Heuristic by relaxed plan extraction Empirical evaluation Conclusion and future work

Action Representation Similar to PDDL2.1 Level 3: –Actions have non-uniform durations and may consume resources –Preconditions are true at start point or hold true for the action duration. –Effects at start or end points. Load(package,truck,place) At(package,place)  At(package,place) At(truck,place) In(package,truck)

The (Relaxed) Temporal PG Tempe Phoenix Los Angeles Drive-car(Tempe,LA) Heli(T,P) Shuttle(T,P) Airplane(P,LA) t = 0t = 0.5t = 1t = 1.5 t = 10

Time-sensitive Cost Function Standard (Temporal) planning graph (TPG) shows the time-related estimates e.g. earliest time to achieve fact, or to execute action TPG does not show the cost estimates to achieve facts or execute actions Tempe Phoenix L.A Shuttle(Tempe,Phx): Cost: $20; Time: 1.0 hour Helicopter(Tempe,Phx): Cost: $100; Time: 0.5 hour Car(Tempe,LA): Cost: $100; Time: 10 hour Airplane(Phx,LA): Cost: $200; Time: 1.0 hour cost time $300 $220 $100  Drive-car(Tempe,LA) Heli(T,P) Shuttle(T,P) Airplane(P,LA) t = 0t = 0.5t = 1t = 1.5 t = 10

Estimating the Cost Function Tempe Phoenix L.A time $300 $220 $100  t = 1.5 t = 10 Shuttle(Tempe,Phx): Cost: $20; Time: 1.0 hour Helicopter(Tempe,Phx): Cost: $100; Time: 0.5 hour Car(Tempe,LA): Cost: $100; Time: 10 hour Airplane(Phx,LA): Cost: $200; Time: 1.0 hour 1 Drive-car(Tempe,LA) Hel(T,P) Shuttle(T,P) t = 0 Airplane(P,LA) t = t = 1 Cost(At(LA))Cost(At(Phx)) = Cost(Flight(Phx,LA)) Airplane(P,LA) t = 2.0 $20

Cost Propagation Issues: –At a given time point, each fact is supported by multiple actions –Each action has more than one precondition Propagation rules: –Cost(f,t) = min {Cost(A,t) : f  Effect(A)} –Cost(A,t) = Aggregate(Cost(f,t): f  Pre(A)) Sum-propagation:  Cost(f,t) Max-propagation: Max {Cost(f,t)} Combination: 0.5  Cost(f,t) Max {Cost(f,t)}

Termination Criteria Deadline Termination: Terminate at time point t if: –  goal G: Dealine(G)  t –  goal G: (Dealine(G) < t)  (Cost(G,t) =  Fix-point Termination: Terminate at time point t where we can not improve the cost of any proposition. K-lookahead approximation: At t where Cost(g,t) < , repeat the process of applying (set) of actions that can improve the cost functions k times. cost time $300 $220 $100  Drive-car(Tempe,LA) H(T,P) Shuttle(T,P) Plane(P,LA) t = t = 10 Earliest time point Cheapest cost

Heuristic estimation using the cost functions If the objective function is to minimize time: h = t 0 If the objective function is to minimize cost: h = CostAggregate(G, t  ) If the objective function is the function of both time and cost O = f(time,cost) then: h = min f(t,Cost(G,t)) s.t. t 0  t  t  Eg: f(time,cost) = 100.makespan + Cost then h = 100x at t 0  t = 2  t  time cost 0 t 0 =1.52t  = 10 $300 $220 $100  Cost(At(LA)) Earliest achieve time: t 0 = 1.5 Lowest cost time: t  = 10 The cost functions have information to track both temporal and cost metric of the plan, and their inter-dependent relations !!!

Heuristic estimation by extracting the relaxed plan Relaxed plan (Hoffman) satisfies all the goals ignoring the negative interaction: –Take into account positive interaction –Base set of actions for possible adjustment according to neglected (relaxed) information (e.g. negative interaction, resource usage etc.)  Need to find a good relaxed plan (among multiple ones) according to the objective function

Heuristic estimation by extracting the relaxed plan Initially supported facts: SF = Init state Initial goals: G = Init goals \ SF Traverse backward searching for actions supporting all the goals. When A is added to the relaxed plan RP, then: SF = SF  Effects(A) G = (G  Precond(A)) \ Effects If the objective function is f(time,cost), then A is selected such that: f(t(RP+A),C(RP+A)) + f(t(G new ),C(G new )) is minimal (G new = (G  Precond(A)) \ Effects) When A is added, using mutex to set orders between A and actions in RP so that less number of causal constraints are violated time cost 0 t 0 =1.52t  = 10 $300 $220 $100  Tempe Phoenix L.A f(t,c) = 100.makespan + Cost

Heuristic estimation by extracting the relaxed plan General Alg.: Traverse backward searching for actions supporting all the goals. When A is added to the relaxed plan RP, then: Supported Fact = SF  Effects(A) Goals = SF \ (G  Precond(A)) Temporal Planning with Cost: If the objective function is f(time,cost), then A is selected such that: f(t(RP+A),C(RP+A)) + f(t(G new ),C(G new )) is minimal (G new = (G  Precond(A)) \ Effects) Finally, using mutex to set orders between A and actions in RP so that less number of causal constraints are violated time cost 0 t 0 =1.52t  = 10 $300 $220 $100  Tempe Phoenix L.A f(t,c) = 100.makespan + Cost

Empirical evaluation Objective: –Demonstrate that metric temporal planner armed with our approach is able to produce plans that satisfy a variety of cost/makespan tradeoff. Testing problems:  Randomly generated logistics problems from TP4 (Hasslum&Geffner) Load/unload(package,location): Cost = 1; Duration = 1; Drive-inter-city(location1,location2): Cost = 4.0; Duration = 12.0; Flight(airport1,airport2): Cost = 15.0; Duration = 3.0; Drive-intra-city(location1,location2,city): Cost = 2.0; Duration = 2.0;

Empirical Results Results over 20 randomly generated temporal logistics problems involve moving 4 packages between different locations in 3 cities: O = f(time,cost) = .Makespan + (1-  ).TotalCost

Empirical Results (cont.) Higher look-ahead option generally produces better results in term of solving times and quality Relaxed plan heuristic is generally more informative than the direct plan heuristic

Related Work TGP, TP4 aim at makespan optimization (do not consider cost) MO-GRT does multi-criteria search, but does not exploit the inter-dependent relations between them. ASPEN (JPL) uses the iterative repairing technique to improve multi-dimensional plan quality

Conclusion Introduced the time-sensitive cost functions to guide the heuristic search according to the objective functions involving both time (makespan) and monetary action cost: –Propagating cost function while building the temporal planning graph –Extract the heuristic values using the cost function –Preliminary experiment result with Sapa showing the utilities of the time-sensitive cost functions

Future Work Experiments with domains and problems from the planning competition Improving the cost function by better propagation rules, mutex information when building the temporal planning graph (TGP approach) Heuristics for tracking other types of planning qualities such as execution flexibility Multi-objective search involving non-combinable criteria