Creating an Efficient Translation from PDDL to DEL

Silm, Paul (2021) Creating an Efficient Translation from PDDL to DEL. Bachelor's Thesis, Artificial Intelligence.

Preview	Text bAI_2021_SilmP.pdf Download (349kB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (127kB)

Abstract

Automated planning programs are used to solve planning problems using search algorithms. Most planners accept the Planning Domain Definition Language (PDDL) as an input language. PDDL uses syntactic approach, which works quite well in most scenarios. However it is not well suited to deal with knowledge of agents, especially knowledge about knowledge. Meta-knowledge can be more intuitively represented with a semantic approach, such as in Dynamic Epistemic Logic (DEL), but the syntax is quite different from that in PDDL, making conversion from PDDL to DEL difficult. Here we present an automatic translation from the PDDL extension MAEPL (Multi-Agent Epistemic Planning Language) to DEL. The translation is written in Haskell, converting plain MAEPL descriptions to the existing data structures used by the Symbolic Model Checker SMCDEL. Additionally our tool checks input for semantic consistency and reports any errors to the user. Examples are used to demonstrate that the translation is able to convert any semantically valid MAEPL problem into DEL. While the translation results in an optimal branching factor for the planning problems' search trees, the search space contains many actions that are never taken. We discuss some methods to increase efficiency and other ideas for future work.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Gattinger, B.R.M.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	16 Jul 2021 09:24
Last Modified:	26 Jul 2021 09:38
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/24907

Actions (login required)

View Item