Learning from situated experiences for a contextual planning guidance

Abstract : This paper presents AgLOTOS as an algebraic language dedicated to the specification of agent plans in ambient systems. AgLOTOS offers two levels of plans: elementary plans which are composed to produce an intention plan; The intention plans which are, in turn, composed to build an agent plan. The composition relies on several operators such as alternative and concurrency. Consequently, the plans can be built automatically as a system of concurrent processes. At the execution level, our approach helps the agent to select an optimal plan preserving the consistency of its intentions. The selection is based on an original and formal construction called contextual planning system (CPS), which presents the potential paths with the associated contexts while removing the inconsistent options. Finally, our CPS is improved by using past-experiences for a better guidance of the agent.
Liste complète des métadonnées

Contributeur : Ahmed Chawki Chaouche <>
Soumis le : dimanche 12 février 2017 - 21:31:23
Dernière modification le : samedi 15 décembre 2018 - 01:31:20
Document(s) archivé(s) le : samedi 13 mai 2017 - 12:21:46


Fichiers produits par l'(les) auteur(s)




Ahmed Chawki Chaouche, Amal El Fallah Seghrouchni, Jean-Michel Ilié, Djamel Eddine Saïdouni. Learning from situated experiences for a contextual planning guidance. Journal of Ambient Intelligence and Humanized Computing, Springer, 2016, 7 (4), pp.555-566. 〈http://link.springer.com/article/10.1007%2Fs12652-016-0342-y〉. 〈10.1007/s12652-016-0342-y〉. 〈hal-01366789〉



Consultations de la notice


Téléchargements de fichiers