Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

SibylSat: Using SAT as an Oracle to Perform a Greedy Search on TOHTN Planning

Gaspard Quenard, Damier Pellier, Humbert Fiorino

TL;DR

SibylSat is presented, a novel SAT-based method designed to efficiently solve totally-ordered HTN problems (TOHTN) that adopts a greedy search approach, enabling it to identify promising decompositions for expansion.

Abstract

This paper presents SibylSat, a novel SAT-based method designed to efficiently solve totally-ordered HTN problems (TOHTN). In contrast to prevailing SAT-based HTN planners that employ a breadth-first search strategy, SibylSat adopts a greedy search approach, enabling it to identify promising decompositions for expansion. The selection process is facilitated by a heuristic derived from solving a relaxed problem, which is also expressed as a SAT problem. Our experimental evaluations demonstrate that SibylSat outperforms existing SAT-based TOHTN approaches in terms of both runtime and plan quality on most of the IPC benchmarks, while also solving a larger number of problems.

SibylSat: Using SAT as an Oracle to Perform a Greedy Search on TOHTN Planning

TL;DR

SibylSat is presented, a novel SAT-based method designed to efficiently solve totally-ordered HTN problems (TOHTN) that adopts a greedy search approach, enabling it to identify promising decompositions for expansion.

Abstract

This paper presents SibylSat, a novel SAT-based method designed to efficiently solve totally-ordered HTN problems (TOHTN). In contrast to prevailing SAT-based HTN planners that employ a breadth-first search strategy, SibylSat adopts a greedy search approach, enabling it to identify promising decompositions for expansion. The selection process is facilitated by a heuristic derived from solving a relaxed problem, which is also expressed as a SAT problem. Our experimental evaluations demonstrate that SibylSat outperforms existing SAT-based TOHTN approaches in terms of both runtime and plan quality on most of the IPC benchmarks, while also solving a larger number of problems.

Paper Structure

This paper contains 21 sections, 6 equations, 3 figures, 1 table, 1 algorithm.

Table of Contents

  1. HTN Planning Problem
  2. Task, Action, Methods, and Task Networks
  3. Planning Problem and Solution
  4. Path Decomposition Tree and SAT planners
  5. SibylSat planner
  6. Planning approach
  7. Example
  8. Searching for solution DT in PDT
  9. Expansion of the PDT
  10. Searching for solution in relaxed PDT
  11. Expanding PDT leafs
  12. Ensuring completeness for recursive domains
  13. Evaluation
  14. Planners
  15. Evaluation Metrics
  16. ...and 6 more sections

Figures (3)

  • Figure 1: Example showing a PDT at some level of decomposition for a problem $P=(L, C , O ,M , c_I, s_I, g)$, where $T_i \in C, M_i \in M \text{ and } A_i \in O$. We see that this PDT does not contain all the DTs because the abstract tasks $T_2$ and $T_8$ are undeveloped. A potential solution DT is highlighted in grey.
  • Figure 2: Toy world state.
  • Figure 3: Comparison of the number of methods developed by SibylSat and Lilotane. A different marker is used for each benchmark.

Theorems & Definitions (4)

  • Definition 1: TOHTN Planning problem
  • Definition 2: Decomposition tree
  • Definition 3: Decomposition Tree solution
  • Definition 4: Path Decomposition Tree