Definition, Responsibilities and Models of the Environment
Danny Weyns, Katholieke Universiteit Leuven, Belgium
In this talk we zoom in on the role and responsibilities of the environment in multiagent systems. We propose a definition of environment, and we discuss basic responsibilities of the environment. Subsequently, we propose a reference model for the environment that describes a functional decomposition of the environment with dataflow among elements. We conclude the talk with an overview of a number of challenging domains in multiagent systems in which environmental aspects play a prominent role. For these domains, the environment provides a unified abstraction to encapsulate the non-agent aspects.
Mechanisms and Opportunities of Environments in MASs
Eric Platon, LIP6, Paris, France - University of Sokendai, Tokyo, Japan
Mechanisms of environments in multiagent systems pertain to models of 'wheels and gears' that link responsibilities of the environment to their engineering. Opportunities pertain to potential assembling of mechanisms into applications. The identification of mechanisms and their opportunities is therefore a central endeavor in the comprehensive study of the environment.
In this talk, we discuss a preliminary taxonomy of environment mechanisms, elaborated from the experience and literature related to environments in the past few years (through e.g. E4MAS and TFGs). We then relate the taxonomy to application opportunities of environments, and we attempt to determine a profile of applications that lends itself to exploit the environment in the design of a multiagent system solution.
Towards an Infrastructure for Artifacts
Mirko Viroli, University of Bologna, Cesena, Italy
The notion of artifact has been introduced as a building block to uniformly understand, design and engineer multiagent system environments for cognitive agents. Artifacts have been shown useful in modelling different environment responsibilities, including organization, coordination, and resource management.
A remarkable step in the development of this research and in its introduction in multiagent system practice is to devise an infrastructure for artifacts, supporting the artifact notion and all the necessary mechanisms to fruitfully exploit it independently of the agent model.
In this talk, we discuss the main issues behind the design and implementation of one such infrastructure, and we sketch a roadmap for future works.
Multiagent Modeling and Simulation: the Role of the Environment
Giuseppe Vizzari, University of Milan, Italy
Alexander Helleboogh, Katholieke Universiteit Leuven, Belgium
Agent Based Modeling represents an approach to model and simulate systems made up of a set of autonomously interacting entities living in an environment. The need to consider the simulated environment in defining a model of the simulated system is generally apparent and recognized. The general goal of the presentation is to highlight the possibilities offered by considering the environment as a first class abstraction, not only in the modeling phase but also in the design and development of the simulator. In the presentation, we discuss: (1) the main concerns of the environment in agent based modeling and simulation (e.g. management of time, space, perception, action); and (2) a first proposal for an abstract model of the environment. The goal of the model is to provide support for the design and development of environments for multiagent system-based simulation systems, by bringing together the various concerns and making their interrelationships explicit.
Location Scope: A Location-based Event Model for Pervasive Environments
Kazutaka Matsuzaki, University of Tokyo, Japan
For a location-aware agent, events about the current location are essential information. Events are mainly issued by the environment. One challenge is to model events that ensure the portability of location-aware agent code. However, existing location models do not seem to meet the requirements of this challenge.
In this talk, we introduce a new event model named "Location Scope" that relies on two main concepts: (1) a location model based on an abstract topology for portability of the agent; (2) a development methodology based on Aspect-Oriented Software Development (AOSD) which also contributes to the portability of the agent.
