Yves Demazeau

From Analysis to Deployment: A Multi-agent Platform Survey

This paper presents a survey on multi-agent platforms, with a particular focus on methodology. It presents the four stages of the construction of a multi-agent system and derives from these stages several criteria for comparison. These criteria are then used to evaluate four selected platforms, and lead to a discussion on the future of multiagent platforms.

Elements about the Emergence Issue: A Survey of Emergence Definitions

Emergence, a concept that first appeared in philosophy, has been widely explored in the domain of complex systems and is sometimes considered to be the key ingredient that makes ‘complex systems’ ‘complex’. Our goal in this paper is to give a broad survey of emergence definitions, to extract a shared definition structure and to discuss some of the remaining issues. We do not know of any comparable surveys about the emergence concept. For this presentation, we start from a broadly applicable approach and finish with more specific propositions. We first present five selected works with a short analysis of each. We then propose a merged analysis in which we isolate a common structure through all definitions but also what we think needs further research. Finally, we briefly describe some perspectives about the emergence engine idea also referred to as emergent engineering.

Bayesian dynamic trust model

In this paper we propose a Bayesian dynamic trust model based on Castelfranchi and Falcones works, thanks to which we can determine the agents trust in another agent. Trust is not only a static mental state: it changes over time. Either some new observations are perceived and modify the trust level, or no observation is perceived and so trust is eroded. Our model takes into account these dynamic aspects by a Bayesian Kalman filter. We present, experiment and discuss our formalism compared with others models. The results obtained with our model are relevant and account for the particular dynamic aspects of trust.

ASIC: An Architecture for Social and Individual Control and its Application to Computer Vision

Most of the work done until now in computational vision deals with the construction of independent visual processing methods that implement a specific subtask of the whole process. In this paper, we are proposing a control architecture, Asic, based on a multi-agent approach. It has been used to build the Mavi system which integrates different visual modules and makes them cooperate. We demonstrate its use to study local and global control in a vision system.

7th International Conference on Practical Applications of Agents and Multi-Agent Systems (PAAMS'09)

PAAMS, the International Conference on Practical Applications of Agents and Multi-Agent Systems is an evolution of the International Workshop on Practical Applications of Agents and Multi-Agent Systems. PAAMS is an international yearly tribune to present, to discuss, and to disseminate the latest developments and the most important outcomes related to real-world applications. It provides a unique opportunity to bring multi-disciplinary experts, academics and practitioners together to exchange their experience in the development of Agents and Multi-Agent Systems. This volume presents the papers that have been accepted for the 2009 edition. These articles capture the most innovative results and this year s trends: Assisted Cognition, E-Commerce, Grid Computing, Human Modelling, Information Systems, Knowledge Management, Agent-Based Simulation, Software Development, Transports, Trust and Security. Each paper has been reviewed by three different reviewers, from an international committee composed of 64 members from 20 different countries. From the 92 submissions received, 35 were selected for full presentation at the conference, and 26 were accepted as posters.

SIGMA: application of multi-agent systems to cartographic generalization

In this article we present the approach and the model used in the SIGMA (SemI automated Generalization using Multi-Agent systems) project, which is being developed in collaboration with the COGIT group of IGN (the French National Geographic Institute) and the LAMA group of the IGA laboratory (the Alpine Geographic Institute).

Dynamical Control in Large-Scale Material Handling Systems through Agent Technology

Delayed arrivals, missing tag codes, flight changes, break-downs, etc. are some of the factors, which make the environment of airport baggage handling systems (BHS) extremely dynamic. Pre-scheduling and optimization is not an option, as identity, destination, and the order of items are unknown until they enter the system, which raises the requirement of dynamic control. The DECIDE project focuses on using multi-agent technologies as generic software components to replace traditional system-specific, centralized control software of production systems. This paper presents an application of multi-agent technologies to be used in large-scale material handling systems, such as BHS. To increase understanding and flexibility, standardized interaction patterns from the FIPA organization are used in combination with ontologies, which defines a vocabulary that enables meta-level communication. A mix of coordinating shop floor agents and mediator agents establish an agent platform capable of coping with the challenges. The investigated BHS problem is part of a larger research project, DECIDE, applying multi-agent controls in production systems.

Vowels co-ordination model

In Multi-Agent Systems, co-ordination plays an important role and addresses a hard problem when taken at distributed and dynamic way. Up to now, several approaches embark solving mechanisms upon agent, interaction, organisation or environment levels indistinctly. However, in a Multi-Agent Oriented Programming perspective, we need to consider the dynamics at every MAS component. In this article we address this MAOP view through a co-ordination model based on the Vowels Paradigm, which offers a componential view of a MAS and a development platform. So, our model takes this decomposition method through a cognitive approach by using multi-agent plan relations and social aspects by means of social dependence. Differently of classical component focused models, we orient our approach to every component of the MAS by addressing some co-ordination approaches into an integrated co-ordination model. This model is illustrated through the RoboCup application in order to evaluate these co-ordination ideas.

An agent-based extensible climate control system for sustainable greenhouse production

The slow adoption pace of new control strategies for sustainable greenhouse climate control by industrial growers, is mainly due to the complexity of identifying and resolving potentially conflicting climate control requirements. In this paper, we present a multi-agent-based climate control system that allows new control strategies to be adopted without any need to identify or resolve conflicts beforehand. This is achieved by representing the climate control requirements as separate agents. Identifying and solving conflicts then becomes a negotiation problem among agents sharing the same controlled environment. Negotiation is done using a novel multi-objective negotiation protocol that uses a generic algorithm to find an optimized solution within the search space. The multi-agent-based control system has been empirically evaluated in an ornamental floriculture research facility in Denmark. The evaluation showed that it is realistic to implement the climate control requirements as individual agents, thereby opening greenhouse climate control systems for integration of independently produced control strategies.

Designing Organized Agents for Cooperation with Real Time Constaints

The aim of this paper is to present our approach for designing Multi-Agent Systems in the context of collective robotics and more generally in the context of real time distributed artificial intelligence applications. The paper presents an agent model (ASTRO) especially adapted to a real time context and shows how the cooperation can be achieved with this model by integrating external organizations and interactions. A design methodology is introduced to build agents using social knowledge (interaction and organization). A platform is presented including software development tools supporting the approach.