Reproduce “Temporal Numeric Planning with Patterns” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Matteo Cardellini, Enrico Giunchiglia

Researchers from DIBRIS, Università degli Studi di Genova, extended their Symbolic Pattern Planning (SPP) framework to address temporal numeric planning problems. Their new PATTY_T planner, utilizing a novel SMT encoding, demonstrated superior problem coverage on benchmark domains and consistently found plans with lower abstract step counts than existing symbolic planners.

Key insights

  • Pattern-based SMT encoding leads to significantly lower planning 'bounds' (fewer abstract steps or pattern applications), resulting in more compact plan representations and smaller SMT instances.
  • The proposed `Π_≺` SMT encoding is formally proven to be both correct (any SMT model is a valid plan) and complete (a plan will be found if one exists) for temporal numeric planning.
  • Aggregating eligible durative actions, especially those with linear numeric effects, within a pattern reduces the number of logical variables and clauses, improving solver efficiency.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.