Planning Graph as a (Dynamic) CSP: Exploiting EBL, DDB and other CSP Search Techniques in Graphplan

@article{Kambhampati2000PlanningGA,
  title={Planning Graph as a (Dynamic) CSP: Exploiting EBL, DDB and other CSP Search Techniques in Graphplan},
  author={S. Kambhampati},
  journal={J. Artif. Intell. Res.},
  year={2000},
  volume={12},
  pages={1-34}
}
  • S. Kambhampati
  • Published 1 February 2000
  • Computer Science
  • J. Artif. Intell. Res.
This paper reviews the connections between Graphplan's planning-graph and the dynamic constraint satisfaction problem and motivates the need for adapting CSP search techniques to the Graphplan algorithm. It then describes how explanation based learning, dependency directed backtracking, dynamic variable ordering, forward checking, sticky values and random-restart search strategies can be adapted to Graphplan. Empirical results are provided to demonstrate that these augmentations improve… 

Figures and Tables from this paper

Exploiting CSP Techniques in Temporal Graphplan

This paper pays attention to a more complex planning problem - temporal planning problem under the Graphplan framework and adopt CSP search techniques in the temporal planning graph, which make the process of solution extraction more efficient.

Distance-Based Goal-Ordering Heuristics for Graphplan

This work discusses the shortcomings of known variable and value ordering strategies for Graphplan's backward search phase, and proposes a novel strategy that is based on a notion of the difficulty of achieving the corresponding subgoal, based on the structure of the planning graph itself.

Planning as Propositional CSP: From Walksat to Local Search Techniques for Action Graphs

This paper presents a framework for solving this class of propositional CSPs using local search in planning graphs, and shows that LPG is significantly faster than Blackbox and other planners based on planning graphs.

Constraint Based Reasoning over Mutex Relations in Planning Graphs during Search

This paper model the problem of finding supports for a sub-goal which arises during the search of the GraphPlan algorithm as a constraint satisfaction problem in which arcconsistency or singleton arc-consistency is maintained.

Constraint Based Reasoning over Mutex Relations in Graphplan Algorithm

It is theoretically shown in this paper that the variant of planning graph forbids strictly more situations than it is forbidden by the standard planning graphs.

New Results about LCGP, a Least Committed GraphPlan

The planner LCGP is improved using a domain-independent heuristic for variable and domain ordering, and compared with HSP-R, a very efficient non-optimal sequential planner, based on an heuristic backward state space search.

An Extension to Least Committed Graphplan for Handling Conditional Effects

A new algorithm is provided that extends Least Committed Graphplan to handling conditional effects, and employs the sub-mutex relations between action components, which makes the construction of planning graph much faster.

Domain-independent temporal planning in a planning-graph-based approach

A temporal planning approach that combines the principles of Graphplan and TGP, and uses the information calculated in the planning graph to deal with a model of actions that include local conditions and effects is presented.

Investigating the Effect of Relevance and Reachability Constraints on SAT Encodings of Planning

Two types of constraints that can be propagated in backward direction, and carry relevance information are developed: Backward mutex (bmutex) and Bmutex constraints, which attempt to capture the idea that two actions (or propositions) are never relevant together.
...

References

SHOWING 1-10 OF 77 REFERENCES

Distance-Based Goal-Ordering Heuristics for Graphplan

This work discusses the shortcomings of known variable and value ordering strategies for Graphplan's backward search phase, and proposes a novel strategy that is based on a notion of the difficulty of achieving the corresponding subgoal, based on the structure of the planning graph itself.

Improving Graphplan's Search with EBL & DDB Techniques

I highlight some inefficiencies of Graphplan's backward search algorithm, and describe how these can be eliminated by adding explanation-based learning and dependency-directed backtracking

Understanding and Extending Graphplan

A reconstruction of Blum and Furst's Graphplan algorithm is provided, and how the backward search phase of Graphplan can be improved by applying techniques from CSP literature, and by teasing apart planning and scheduling phases in Graphplan.

Extracting Effective and Admissible State Space Heuristics from the Planning Graph

The planning graph structure that Graphplan builds in polynomial time, provides a rich substrate for deriving more effective heuristics for state space planners, and the mutex information in the planning graph captures exactly this interaction information.

Unifying SAT-based and Graph-based Planning

It is shown that STRIPS problems can be directly translated into SAT and efficiently solved using new randomized systematic solvers and that polynomialtime SAT simplification algorithms applied to the encoded problem instances are a powerful complement to the "mutex" propagation algorithm that works directly on the plan graph.

Dynamic Variable Ordering in CSPs

This work provides an implementation methodology for adding DVO to an arbitrary tree-search algorithm and investigates the popular reordering heuristic of next instantiating the variable with the minimum remaining values (MRV).

Fast Planning Through Planning Graph Analysis

Investigating the Effect of Relevance and Reachability Constraints on SAT Encodings of Planning

Two types of constraints that can be propagated in backward direction, and carry relevance information are developed: Backward mutex (bmutex) and Bmutex constraints, which attempt to capture the idea that two actions (or propositions) are never relevant together.

Exploiting Symmetry in the Planning graph via Explanation-Guided Search

The method consists of maintaining a pilot explanation structure capturing the failures encountered at previous levels of the search, and using it in an intelligent way to guide the search at the newer levels.

Nogood Recording for static and dynamic constraint satisfaction problems

A new class of constraint recording algorithms called Nogood Recording is proposed that may be used for solving both static and dynamic CSPs and offers an interesting compromise, polynomially bounded in space, between an ATMS-like approach and the usual static constraint satisfaction algorithms.
...