Patty

Patty solves numeric planning problems as logic: it encodes them for an SMT solver, using patterns of actions to keep the encoding small.

Authors

Matteo Cardellini, Enrico Giunchiglia, Marco Maratea

Institutions

At publication: Politecnico di Torino, University of Genova, University of Calabria

Kind of planning

Numeric and classical

Source

github.com/matteocarde/patty

How it works

Patty belongs to the Planning as Satisfiability family. Given a problem, a bound and a pattern, "defined as an arbitrary sequence of actions", it encodes the search for a plan as a formula with fewer variables or clauses than earlier encodings, and hands that to a solver. Its IPC 2026 version refines patterns as it goes, and its satisficing versions remove redundant actions from the plan afterwards.

What it handles

Numeric conditions and effects. It does not handle derived predicates or durative actions. We could not confirm several other features, marked as unverified in the feature comparison.

In the competitions

Patty was entered in the IPC 2026 numeric track by Matteo Cardellini and Enrico Giunchiglia, and one of its configurations took a medal in the simple numeric satisficing category, according to the published results.

Results in our benchmarks

Patty a5001cc: problems solved with a plan the VAL validator accepts, within 60 seconds and 4 GB of memory on one core. These are our own runs, not competition results, and a planner written for other goals or for a different kind of problem can score low here while being the right tool for its purpose. How we measured.

Papers

1. Matteo Cardellini, Enrico Giunchiglia, Marco Maratea. Symbolic Numeric Planning with Patterns. Proceedings of AAAI 38(18), 20070-20077, 2024.
2. Matteo Cardellini, Enrico Giunchiglia, Marco Maratea. Symbolic pattern planning. Artificial Intelligence 352, 104482, 2026.
3. Matteo Cardellini, Enrico Giunchiglia. PATTY: The Symbolic Pattern Planner. Planner abstract, IPC 2026 numeric track.