PPDDL1.0: An extension to PDDL for expressing planning domains with probabilistic effects

Abstract

We desribe a variation of the planning domain definition language, PDDL, that permits the modeling of probabilistic planning problems with rewards. This language, PPDDL1.0, was used as the input language for the probabilistic track of the 4th International Planning Competition. We provide the complete syntax for PPDDL1.0 and give a semantics of PPDDL1.0 planning problems in terms of Markov decision processes.

Sample citation

Håkan L. S. Younes and Michael L. Littman. 2004. PPDDL1.0: An extension to PDDL for expressing planning domains with probabilistic effects. Technical Report CMU-CS-04-167, School of Computer Science, Carnegie Mellon University, Pittsburgh, Pennsylvania.

Full paper (23 pages, 19 references)