Finite-state controllers are a compact and effective plan representation for agent widely used in AI. In this paper, we propose a generic framework and related solver for synthesizing bounded finite-state controllers, and show its instantiations to three different applications, including generalized planning, planning programs and service composition under partial observability and controllability. We show that our generic solver is sound and complete, and amenable to heuristics that take into account the structure of the specific target instantiation. Experiments show that instantiations of our solver to the problems above often outperform tailored approaches in the literature. This suggests that our proposal is a promising base point for future research on finite-state controller synthesis. Copyright © 2013, Association for the Advancement of Artificial Intelligence. All rights reserved.
A generic technique for synthesizing bounded finite-state controllers / Yuxiao, Hu; DE GIACOMO, Giuseppe. - STAMPA. - (2013), pp. 109-116. (Intervento presentato al convegno 23rd International Conference on Automated Planning and Scheduling, ICAPS 2013 tenutosi a Rome nel 10 June 2013 through 14 June 2013).
A generic technique for synthesizing bounded finite-state controllers
DE GIACOMO, Giuseppe
2013
Abstract
Finite-state controllers are a compact and effective plan representation for agent widely used in AI. In this paper, we propose a generic framework and related solver for synthesizing bounded finite-state controllers, and show its instantiations to three different applications, including generalized planning, planning programs and service composition under partial observability and controllability. We show that our generic solver is sound and complete, and amenable to heuristics that take into account the structure of the specific target instantiation. Experiments show that instantiations of our solver to the problems above often outperform tailored approaches in the literature. This suggests that our proposal is a promising base point for future research on finite-state controller synthesis. Copyright © 2013, Association for the Advancement of Artificial Intelligence. All rights reserved.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
