Gaële Simon

Goal decomposition tree: an agent model to generate a validated agent behaviour

This paper deals with a goal-oriented agent model called Goal Decomposition Tree (GDT) allowing both to specify and validate the behaviour of an agent. This work takes place in a global framework whose goal is to define a process allowing to start from a problem specification to obtain a validated implementation of a corresponding MAS. The GDT model has been used to specify a prey-predator system which has been verified this way.

Implementing validated agents behaviours with automata based on goal decomposition trees

In order to provide an effective tool allowing to implement validated agents behaviours, this paper first presents a Goal Decomposition Tree (GDT), a model to specify behaviours both in procedural and declarative ways. A GDT allows the designer to verify the specified behaviour. This model is then used to generate a behaviour automaton using automata composition patterns associated to operators used in the tree. This process allows to obtain a finite expression representing all valid behaviours of agents of a MAS.

A Methodology to Solve Optimisation Problems with MAS Application to the Graph Colouring Problem

Developing multi-agent systems may be a rather difficult task. Having confidence in the result is still more difficult. In this article, we describe a methodology that helps in this task. This methodology is dedicated to global optimization problems that can be solved combining local constraints. We developed CASE tools to support this methodology which are also presented. Finally, we show how this methodology has been successfully used to develop a multi-agent system for the graph colouring problem.

Specifying and verifying holonic agents with GDT4MAS

This paper describes how specific holonic multi-agent systems can be specified and how their correctness can be proven with an extended version of the GDT4MAS model. This model allows the specification of multi-agent systems and the verification of their correctness with theorem proving techniques. Introducing holonic agents in this model allows the enhancement of its expressiveness. Moreover, the proof system associated to this model can be easily extended in order to prove the correctness of multi-agent systems using such agents. The paper first describes the initial GDT4MAS model. Then the need for some kinds of holonic agents and their proposed specification based on specific decomposition operators are presented. It is followed by a focus on how the proof system can be adapted to prove the correctness of the behaviour of these new agents. Last but not least, all these proposals are illustrated on a case study.

A Dynamic Clustering Algorithm for Mobile Objects

In this paper, a multiagent algorithm for dynamic clustering is presented. This kind of clustering is intended to manage mobile data and so, to be able to continuously adapt the built clusters. First of all, potential applications of this algorithm are presented. Then the specific constraints for this kind of clustering are studied. A multiagent architecture satisfying these constraints is described. It combines an ants algorithm with a cluster agents layer which are executed simultaneously. Finally, the first experimental results of our work are presented.

A New Proof System to Verify GDT Agents

The GDT4MAS model is dedicated to the formal specification of agents and multiagents systems. It has been presented in previous articles [7, 6]. The proof mechanism relies on proof schemas that generate proof obligations, that is to say first-order formulae that can be proven (in most cases) by an automatic prover (such as PVS). In this article, we present a new version of proof schemas that increase the number of proofs that can be performed.

Vérification formelle du respect de valeurs morales dans les SMA

L’utilisation croissante d’agents autonomes artificiels dans des secteurs comme le milieu hospitalier ou les transports amene a reflechir au respect de regles morales partagees par tous. Le probleme est d’autant plus crucial que les regles morales informellement validees par tous sont souvent incompatibles les unes avec les autres, et que ce sont souvent des regles ethiques qui amenent l’humain, suivant les circonstances, a privilegier telle ou telle regle morale. Utiliser la preuve pour verifier qu’un agent respecte bien des regles morales et ethiques pourraient aider a accroitre la confiance que nous pouvons avoir en de telles unites logicielles. Dans cet article, nous montrons, en nous appuyant sur une etude de cas, comment, a partir de regles morales a priori contradictoires mais ordonnees par des regles ethiques exprimant un systeme de valeurs, nous parvenons a definir un ensemble de proprietes formelles qui, lorsqu’elles sont etablies par un agent, permettent d’assurer que l’agent en question respecte bien la regle ethique souhaitee.

Using GDT4MAS as a formal support for engineering multi-agents systems

This paper focuses on multi-agent systems engineering process. An assessment of current needs in this domain, based on the analysis of systems already developed, is performed. This assessment shows that the formal verification of MAS is one of these needs. It is then shown how the formal approach GDT4MAS provides answer to many of the other needs. This approach is based on a MAS formal specification associated to a proof process allowing to establish the correctness of properties of the system. The main purpose of this paper is to show that, unlike most other formal approaches for MAS, GDT4MAS can at the same time propose formal aspects making a proof possible and contribute to different general aspects of agent-oriented software engineering, even when formal verification is not a concern.

Side Effects of Agents Are Not Just Random

Side effects are an important characteristic of MAS, and proving them is an interesting issue. They often can be expressed as liveness properties. But there is no system dedicated to this kind of proof. The GDT4MAS framework allows to specify and prove the correctness of multiagent systems. This framework is mainly dedicated to prove safety properties about the system and to prove that agents achieve their goal(s). However, there is no proof principle to prove that agents satisfy liveness properties that are not part of their goal(s). In this article, we propose a proof mechanism that addresses this kind of problem: we show how we can add to GDT4MAS a proof mechanism adapted to prove leads-to properties, a subclass of liveness properties.

Using the SPACE Intermediate Model to Help to Implement Agents

In this article, an intermediate model for specifying agents is described. This model makes easier the transition between a conceptual model (like the BDI one) and a multiagent environment. The SPACE model is specified. An example of application to BDI agents is also presented.