Models and Software

Planning Models

I actively enjoy encoding knowledge planning models, usually in PDDL or PDDL+.

Some of the models I engineered have been used in International Competitions, while others have been used for testing the exploitability of planning in some real-world applications. I also modified well-known benchmark models to test the robustness of state-of-the-art domain-independent planning engines.

All the models can be found in a bitbucket repository I specifically designed, https://bitbucket.org/maurovallati/planning-knowledge-models

Some example models we developed for using PDDL+ planning for in-station train dispatching can be found here: https://github.com/matteocarde/icaps2021

The deployable PDDL+ models for urban traffic control, with small example problems, can be found here: https://github.com/anas-elkouaiti/utc-models-deployable 

Feel free to use the models for your research, or just to play a bit with planning engines.

Software Developed

• PbP, a planning system based on an automatically configurable portfolio of planners (with Alfonso Gerevini and Alessandro Saetti); Winner of the learning track of the 6th International Planning Competition (2008). PbP2, an improved version, was the winner of the learning track of the 7th International Planning Competition (2011).

• MacroSatPlan, SAT-based optimal planner which exploits a predictive model and macros to speedup the SAT solving (with Alfonso Gerevini and Alessandro Saetti).

• ParLPG, a planning system based on the idea of automatically configuring a generic,parameterized planner: LPG. (with Alfonso Gerevini, Alessandro Saetti, Chris Fawcett and Holger Hoos).

• TemPoRal, a portfolio-based temporal planning system developed (in fact, mainly by Isabel Cenamor) for the IPC 2018. It was also the winner, btw.

• ArgSemSAT, developed with Federico Cerutti, is a novel SAT-based approach for solving various tasks in abstract argumentation. It was the runner-up of the 1st International Competition on Computational Models of Argumentation (ICCMA), and the winner of the PR track of the 2nd International Competition on Computational Models of Argumentation (ICCMA).

• fudge, developed with Matthias Thimm and Federico Cerutti, is a lightweight yet extremely performant SAT-based solver for various tasks in abstract argumentation. It was the winner of the ideal track of ICCMA 2021.

• JArgSemSAT, is a portable java-based version of ArgSemSAT.

• AFBenchGen, developed with Federico Cerutti and Massimiliano Giacomin, is a highly configurable generator of challenging Abstract Argumentation frameworks. It is described in this paper.