Authors:
Sabrine Ammar
1
and
Mohamed Tahar Bhiri
2
Affiliations:
1
MIRACL, Faculty of Economics and Management of Sfax, Sfax, Tunisia
;
2
MIRACL, Faculty of Science of Sfax, Sfax, Tunisia
Keyword(s):
Code Generation, Correct by Construction, Event-B, PDDL, Formal Modelling, Refinement Strategy.
Abstract:
In artificial intelligence, the goal of automatic planning is to structure actions in the form of a plan to achieve an expressed goal. The PDDL (Planning Domain Definition Language) was designed to allow the common representation of planning problems during ICAPS (International Conference on Automated Planning and Scheduling) competitions. PDDL has many verification and validation tools allowing the description, resolution and validation of planning problems. But they only allow the reliability of PDDL descriptions a posteriori. In this article, we recommend a rigorous approach coupling Event-B and PDDL favoring obtaining PDDL descriptions deemed correct, a priori, from an ultimate Event-B model. The formal Event-B method allows us to obtain, by successive refinements with mathematical proofs, correct by construction formal models of planning problems. A refinement strategy appropriate to planning problems is, then, proposed. The ultimate Event-B model, correct by construction, is au
tomatically translated into PDDL using our MDE Event-B2PDDL tool. The obtained PDDL description is submitted to efficient planners for generation of correct and efficient plan-solutions.
(More)