Dealing with Dynamicity in Ambient Systems: A Formal Planning Approach.
Intervenant(s) : Jean-Michel ILIE (LIP6 et Univ. Paris 5 Descartes)
Ambient Intelligence (AmI) systems can be viewed as highly dynamical distributed and embedded systems. Face to the practical interests of implementing SMART systems with them, many works in the literature propose AmI implementations from classical languages. In this talk, we demonstrate the interest of a formal-based software architecture, which can originally be used to design and implement AmI systems, taking into account the context-awareness of agents. Based on an agent-centric development, the core of this work is a new algebraic language, namely AgLOTOS, used to write plans and compose them, automatically and on-the-fly, from the current intentions of the agent. The associated semantics interestingly allows to view the plan of an agent as a system of concurrent processes with possible failures in the execution. In this context, we both aim at ensuring the agent consistency on-the-fly and efficiently guiding him, in its next execution. The current experiments show the interest of this approach to build a SMART University Campus.