Agent Development Techniques and Multi-Agent Platforms
Résumé
When talking to industrial partners and trying to convince them to use agent technology, you usually hear complaints about the lack of available tools and of industrial analysis and design methods for doing a decent job. Although agent and multi-agent technology has been with us for many years, to many people developing applications remains a sort of black magic. This results probably from various shortcomings of the available tools. I will try to point at some of the practical problems we encountered during the last 25 years when developing multi-agent systems and to mention directions for improvement based on our experience. In the first part of my talk I will address the following issues: What kind of application? What domain? What purpose? I will distinguish among distributed applications, simulation, and games, giving examples of various systems we developed. In the second part, I will address the following issues: Who is implied? What features do we need? For whom? From the people involved, namely end-users, application developers and platform designers, I will consider requirements from application developers with consequences for the platform designers. Issues like agent communication language, content language, time control, multi-processing, proactivity, representation, agent memory organization, Human-agent interaction are important points to be examined. In the third part of my talk, I will present some of the features that have been implemented in a platform we have been developing, trying to answer some of the issues previously mentioned. I will then conclude and answer questions.