Jacques Ferber

A meta-model for the analysis and design of organizations in multi-agent systems

This paper presents a generic meta-model of multi-agent systems based on organizational concepts such as groups, roles and structures. This model, called AALAADIN, defines a very simple description of coordination and negotiation schemes through multi-agent systems. Aalaadin is a meta-model of artificial organization by which one can build multi-agent systems with different forms of organizations such as market-like and hierarchical organizations. We show that this meta-model allows for agent heterogeneity in languages, applications and architectures. We also introduce the concept of organizational reflection which uses the same conceptual model to describe system level tasks such as remote communication and migration of agents. Finally, we briefly describe a platform, called MADKIT, based on this model. It relies on a minimal agent kernel with platform-level services implemented as agents, groups and roles.

From Agents to Organizations: An Organizational View of Multi-agent Systems

While multi-agent systems seem to provide a good basis for building complex software systems, this paper points out some of the drawbacks of classical agent centered multi-agent systems. To resolve these difficulties we claim that organization centered multi-agent system, or OCMAS for short, may be used. We propose a set of general principles from which true OCMAS may be designed. One of these principles is not to assume anything about the cognitive capabilities of agents. In order to show how OCMAS models may be designed, we propose a very concise and minimal OCMAS model called AGR, for Agent/Group/Role. We propose a set of notations and a methodological framework to help the designer to build MAS using AGR. We then show that it is possible to design multi-agent systems using only OCMAS models.

The MADKIT Agent Platform Architecture

In this paper, we present MadKit (multi-agent development kit), a generic multi-agent platform. This toolkit is based on a organizational model. It uses concepts of groups and roles for agents to manage different agent models and multi-agent systems at the same time, while keeping a global structure. We discuss the architecture of MadKit, based on a minimalist agent kernel decoupled from specific agency models. Basic services like distributed message passing, migration or monitoring are provided by platform agents for maximal flexibility. The componential interface model allows variations in platform appearance and classes of usage.

MadKit: a generic multi-agent platform

This paper describes brielfy MadKit, a generic multi-agent platform. This toolkit is based on a organizational model. It uses concepts of groups and roles for agents to manage different agent models and multi-agent systems at the same time, while keeping a global structure. The platform architecture is based on a minimalist agent kernel decoupled from specific agency models. Basic services like distributed message passing, migration or monitoring are provided by platform agents for maximal flexibility. A componential interface model allows variations in platform appearance and classes of usage.

Organization models and behavioural requirements specification for multi-agent systems

The main question addressed in this paper is how requirements on the dynamics of a multi-agent systems and individual agents can be related to the dynamics of high-level concepts given by an organisation model, such as groups, roles within groups, and role interaction.

Integrating tools and infrastructures for generic multi-agent systems

In this paper, we present MadKit/SEdit, an agent infrastructure combined with a generic design tool for multi-agent systems. This toolkit is based on a organizational metaphor to integrate highly heterogeneous agent systems. We explain the principles of MadKit, the underlying agent platform, and show how it can integrate various agent architectures and provides structuration for multiple simultaneous systems and semantics. The architecture, based on a minimal agent runtime, agentified platform services and modular application host, is presented. The SEdit design tool, built itself as a MAS is also discussed. We present its key points in terms of multi-model support, and integration with the infrastructure, from design to maintenance. We illustrate our approach by discussing some consequences of this architecture, and describe our motivation for this design: integration and reuse, organizational patterns, and overall versatility. A summary is given of some key MadKit-based applications to date.

Action selection in an autonomous agent with a hierarchical distributed reactive planning architecture

Action selection is a central issue for robotic autonomous agent design. The action selection mechanism used in an autonomous robot needs to be reactive, but must include planning capacities. It should ennble the coexistence of antinomic goals and constraints, and induce a sufficient adaptivity to cope with circumstances unexpected at design time. To our knowledge, no such mechanism embedding all these properties has been implemented in a robotic application. We propose here an action selection mechanism that fulfill all these requirements and present a principle test of its properties on a simple robot realistic simulator.

Operational Semantics of Multi-agent Organizations

This paper introduces a formal description of the operational semantics of multiagent organizations expressed in the Aalaadin generic model. This formalization is based on the π-calculus and the Chemical Abstract Machine (Cham).

Weak Interaction and Strong Interaction in Agent Based Simulations

This paper addresses the problem of the engineering divergence phenomenon in ABS. This problem is related to the fact that a particular conceptual model may give different outputs according to its implementation. Through two experiments, the paper shows that the implementation of the agents’ interaction is one of the factors that are involved in this phenomenon. The underlying idea of this paper is that this problem can be greatly diminished if the analysis of the conceptual model incorporates some key concepts which are crucial for the implementation. To this end, this work proposes to identify two different classes of interaction: weak interactions and strong interactions.

Generic architecture for multi-AUV cooperation based on a multi-agent reactive organizational approach

Because Autonomous Underwater Vehicles (AUVs) have limited perception and communication capabilities, designing efficient AUV flotillas is challenging. Existing solutions are often strongly related to (1) a specific kind of mission and (2) the nature of the considered AUVs. So, it is difficult to reuse these approaches when switching to another mission context. This paper proposes a generic multi-agent based layered architecture for designing and specifying AUV flotillas at a high level of abstraction, regardless of the AUVs characteristics and skills. To this end, an organizational model is used to ease and regulate interactions between heterogeneous AUVs and combined with a behavioral reactive approach for limiting communication.