Abstract: The advent of multiagent systems has brought together many disciplines and given us a new way to look at intelligent, distributed systems. However, traditional ways of thinking about and designing software do not fit the multiagent paradigm. This paper describes the Multiagent Systems Engineering (MaSE) methodology and agentTool, a tool to support MaSE. MaSE guides a designer from an initial system specification to implementation by guiding the designer through a set of inter-related graphically based system models. The underlying formal syntax and semantics of clearly and unambiguously ties them together as envisioned by MaSE.

Abstract: Distributed Artificial Intelligence (DAI) has existed as a subfield of AI for less than two decades. DAI is concerned with systems that consist of multiple independent entities that interact in a domain. Traditionally, DAI has been divided into two sub-disciplines: Distributed Problem Solving (DPS) focuses on the information management aspects of systems with several components working together towards a common goals Multiagent Systems (MAS) deals with behavior management in collections of several independent entities, or agents. This survey of MAS is intended to serve as an introduction to the field and as an organizational framework. A series of general multiagent scenarios are presented. For each scenario, the issues that arise are described along with a sampling of the techniques that exist to deal with them. The presented techniques are not exhaustive, but they highlight how multiagent systems can be and have been used to build complex systems. When options exist, the techniques presented are biased towards machine learning approaches. Additional opportunities for applying machine learning to MAS are highlighted and robotic soccer is presented as an appropriate test bed for MAS. This survey does not focus exclusively on robotic systems. However, we believe that much of the prior research in non-robotic MAS is relevant to robotic MAS, and we explicitly discuss several robotic MAS, including all of those presented in this issue.

Abstract: The ability to communicate is one of the salient properties of agents. Although a number of agent communication languages (ACLs) have been developed, obtaining a suitable formal semantics for ACLs remains one of the greatest challenges of multiagent systems theory. Previous semantics have largely been mentalistic in their orientation and are based solely on the beliefs and intentions of the participating agents. Such semantics are not suitable for most multiagent applications, which involve autonomous and heterogeneous agents, whose beliefs and intentions cannot be uniformly determined. Accordingly, we present a social semantics for ACLs that gives primacy to the interactions among the agents. Our semantics is based on social commitments and is developed in temporal logic. This semantics, because of its public orientation, is essential to providing a rigorous basis for multiagent protocols.

Abstract: With deregulation sweeping all over electrical systems around the world, transmission planning has undergone dramatic changes during this decade. Centralized cost allocation methods have become obsolete and new procedures are needed to deal with intelligent and self-sufficient players. In this paper, the authors study the allocation of transmission costs in a decentralized manner. For this purpose, they have developed a multi-agent system that is based on a well known cooperative game theory procedure, the kernel. Using their approach, the agents are able to form kernel-stable coalitions and the cost allocation procedure is performed at every step of the kernel algorithm. A six bus example and an IEEE 24 bus case illustrate their model.

Abstract: This chapter contains sections titled: 2.1 Introduction, 2.2 Agent Communications, 2.3 Agent Interaction Protocols, 2.4 Societies of Agents, 2.5 Conclusions, 2.6 Exercises, 2.7 References

Abstract: : Learning behaviors in a multiagent environment are crucial for developing and adapting multiagent systems. Reinforcement learning techniques have addressed this problem for a single agent acting in a stationary environment, which is modeled as a Markov decision process (MDP). But, multiagent environments are inherently non-stationary since the other agents are free to change their behavior as they also learn and adapt. Stochastic games, first studied in the game theory community, are a natural extension of MDPs to include multiple agents. In this paper we contribute a comprehensive presentation of the relevant techniques for solving stochastic games from both the game theory community and reinforcement learning communities. We examine the assumptions and limitations of these algorithms, and identify similarities between these algorithms, single agent reinforcement learners, and basic game theory techniques.

Abstract: This paper introduces the RoboCup-Rescue Simulation Project, a contribution to the disaster mitigation, search and rescue problem. A comprehensive urban disaster simulator is constructed on distributed computers. Heterogeneous intelligent agents such as fire fighters, victims and volunteers conduct search and rescue activities in this virtual disaster world. A real world interface integrates various sensor systems and controllers of infrastructures in the real cities with the virtual world. Real-time simulation is synchronized with actual disasters, computing complex relationship between various damage factors and agent behaviors. A mission-critical man-machine interface provides portability and robustness of disaster mitigation centers, and augmented-reality interfaces for rescue parties in real disasters. It also provides a virtual reality training function for the public. This diverse spectrum of RoboCup-Rescue contributes to the creation of the safer social system.

Abstract: Automated negotiation and coalition formation among self-interested agents are playing an increasingly important role in electronic commerce. Such agents cannot be coordinated by externally imposing their strategies. Instead the interaction protocols have to be designed so that each agent is motivated to follow the strategy that the protocol designer wants it to follow. This paper reviews six component technologies that we have developed for making such interactions less manipulable and more efficient in terms of the computational processes and the outcomes: 1. OCSM-contracts in marginal cost based contracting, 2. leveled commitment contracts, 3. anytime coalition structure generation with worst case guarantees, 4. trading off computation cost against optimization quality within each coalition, 5. distributing search among insincere agents, and 6. unenforced contract execution. Each of these technologies represents a different way of battling self-interest and combinatorial complexity simultaneously. This is a key battle when multi-agent systems move into large-scale open settings.

Abstract: Agent technology derived from Distributed Artificial Intelligence is increasingly being considered for next generation computer-integrated manufacturing systems, to satisfy new requirements for increased integrability, configurability, adaptability, extendibility, agility, and reliability. This paper reviews our previous research on the application of the agent-based technology to intelligent design and manufacturing and describes the current research project MetaMorph II (an agent-based architecture for distributed intelligent design and manufacturing).

Abstract: Since the beginning of multiagent systems research, it has been argued that theories from the realm of the social sciences can be of help when building multiagent systems, and to some extent vice versa. This study sketches the concepts necessary for agents based on social theories. The concepts, viz. agent, autonomy and norms, used in these theories are discussed and defined. The social theories include theories on decision making, various rationality and action models, and the role and modeling of other agents. The sociological debate on the micro-macro problem is analyzed and translated to multiagent research and combined with philosophical theories on sociality. An agent typology is proposed and linked to the communication and modeling possibilities of the agents. Concluding the thesis is a selection of articles presented at workshops and conferences, focusing on the usefulness of norms for artificial adjustable autonomous agents, and two articles on simulation studies in order to develop and test organization theories.

Abstract: Little-JIL, a new language for programming the coordination of agents, is an executable, high-level process programming language with a formal (yet graphical) syntax and rigorously defined operational semantics. Little-JIL is based on two main hypotheses. The first is that the specification of coordination control structures is separable from other process programming language issues. Little-JIL provides a rich set of control structures while relying on separate systems for support in areas such as resource, artifact and agenda management. The second hypothesis is that processes can be executed by agents who know how to perform their tasks but can benefit from coordination support. Accordingly, each step in Little-JIl is assigned to an execution agent (human or automated). These agents are responsible for initiating steps and performing the work associated with them. This approach has so far proven effective in allowing us to clearly and concisely express the agent coordination aspects of a wide variety of software, workflow and other processes.

Abstract: The problem of efficient multirobot coordination has risen to the forefront of robotics research in recent years. Interest in this problem is motivated by the wide range of application domains demanding multirobot solutions. In general, multirobot coordination strategies assume either a centralized approach, where a single robot/agent plans for the group, or a distributed approach, where each robot is responsible for its own planning. Inherent to many centralized approaches are difficulties such as intractable solutions for large groups, sluggish response to changes in the local environment, heavy communication requirements, and brittle systems with single points of failure. The key advantage of centralized approaches is that they can produce globally optimal plans. While most distributed approaches can overcome the obstacles inherent to centralized approaches, they can only produce suboptimal plans. This work explores the development of a market-based architecture that will be inherently distributed, but will also opportunistically form centralized sub-groups to improve efficiency, and thus approach optimality. Robots will be self-interested agents, with the primary goal of maximizing individual profits. The revenue/cost models and rules of engagement will be designed so that maximizing individual profit has the benevolent effect of moving the team toward the globally optimal solution. This architecture will inherit the flexibility of market-based approaches in allowing cooperation and competition to emerge opportunistically. The outlined approach will address the multirobot control problem for autonomous robotic colonies carrying out complex tasks in dynamic environments where it is highly desirable to optimize to whatever extent possible. Future work will develop the core components of a market-based multirobot control-architecture, investigate the use of a negotiation protocol for task distribution, design and implement resource and role management schemes, and apply optimization techniques to improve system performance. The automated robot colonies domain is targeted for implementation and evaluation of the architecture. Portability of the architecture to other application domains will also be illustrated.

Abstract: Scalability is an issue that becomes important when developing practical software agent systems, to perform some of the applications that agent development tools identify. In this paper we address issues such as whether scalability is a theme that can be investigated in isolation to a particular agent development environment or application, and more importantly, whether metrics can be identified to compare the relative performance of multi-agent systems. Scalability is considered from a performance engineering perspective, with the additional constraint of trying to model both mobile and intelligent agents using the same techniques. Petri net based performance models are presented, which can be run on various publicly available simulators. We summarise approaches taken by different segments of the agents community, and use this to identify the disparity in what is termed as 'Scalability'. Our main contribution is to highlight the importance of combining performance engineering with agent oriented design methodologies, to design and build large agent based applications. Content Areas: agent-based software engineering, designing agent systems, lessons learned from deployed agents, multi-agent communication coordination and collaboration, organization

Abstract: Recently many studies have been made on the automatic design of complex systems using evolutionary optimization techniques such as genetic algorithms (GA), evolution strategy (ES), evolutionary programming (EP) and genetic programming (GP). It is generally recognized that these techniques are very useful for optimizing fairly complex systems such as the generation of intelligent behavior sequences of robots. A new method, genetic network programming (GNP), is proposed in order to acquire these behavior sequences efficiently. GNP is composed of plural nodes for agents to execute simple judgment/processing and they are connected with each other to form a network structure. Agents behave according to the contents of the nodes and their connections in GNP. In order to obtain a better structure, the GNP changes itself using evolutionary optimization techniques.

Abstract: The Karma-Teamcore framework focuses on rapidly integrating distributed, heterogeneous agents and tasking them via an abstract team-oriented program. The framework provides wrappers that encapsulate general teamwork reasoning and automatically generate the necessary coordination for robust execution of this abstract program. We describe the Karma-Teamcore framework and present an example of its successful application, namely, the simulated evacuation of civilians stranded in a hostile area.

Abstract: 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.

Abstract: In a multi-agent environment, whether one agent's action is good or not depends on the other agents' actions. In traditional reinforcement learning methods, which assume stationary environments, it is hard to take into account of the other agent's actions which may change due to learning. In this article, we consider a two-agent cooperation problem, and propose a multi-agent reinforcement learning method based on estimation of the other agent's actions. In our learning method, one agent estimates the other agent's action based on the internal model of the other agent. The internal model is acquired by the observation of the other agent's actions. Through experiments, we demonstrate that good cooperative behaviors are achieved by our learning method.

Abstract: The paper is framed to address the preliminary approach towards process oriented collaborative inventory management in supply chains, taking advantage of multi-agent technology in terms of modeling and simulation. Initially, a SCM support model is proposed as a foundation to combine the supply chain processes with the multi-agent system. In succession, a simple PC assembling case is investigated and simulated mainly to validate the SCM support model. As a result, the combination has the potential to make possible a real strategic competitive advantage for the entire supply chain and will enable new forms of business, namely, collaborative inventory management. Accordingly, a theoretical framework of collaborative inventory management is highlighted to refine and extend the SCM support model with the purpose of synchronizing decisions as well as actions.

Abstract: In this paper, we present DIAMS, a system of distributed, collaborative agents to help users access, manage, share and exchange information. A DIAMS personal agent helps its owner find information most relevant to current needs. It provides tools and utilities for users to manage their information repositories with dynamic organization and virtual views. Flexible hierarchical display is integrated with indexed query search-to support effective information access. Automatic indexing methods are employed to support user queries and communication between agents. Contents of a repository are kept in object-oriented storage to facilitate information sharing. Collaboration between users is aided by easy sharing utilities as well as automated information exchange. Matchmaker agents are designed to establish connections between users with similar interests and expertise. DIAMS agents provide needed services for users to share and learn information from one another on the World Wide Web.

Abstract: There is an increasing demand for designers and developers to construct ever larger multi-agent systems. Such systems will be composed of hundreds or even thousands of autonomous agents. Moreover, in open and dynamic environments, the number of agents in the system at any one time will fluctuate significantly. To cope with these twin issues of scalability and variable numbers, we hypothesize that multi-agent systems need to be both self-building (able to determine the most appropriate organizational structure for the system by themselves at run-time) and adaptive (able to change this structure as their environment changes). To evaluate this hypothesis we have implemented such amultiagent system and have applied it to the domain of automated trading. Preliminary results supporting the first part of this hypothesis are presented: adaption and self-organization do indeed make the system better able to cope with large numbers of agents.

Abstract: A method and apparatus for automatically locating sources of semantic error in a multi-agent system based on setup connection tree information, and informing the appropriate agents so that they can avoid using the faulty resources in the future. The setup connection tree model is established based on patterns of agent actions for expressing the logical relationship between available resources in the disjunctive normal form (d.n.f.). A table is used to record different sets of resources for use in the resource selection process. Thus, faulty resources can be located by means of induction. A global database is also maintained for updating information on semantic errors in the system.

Abstract: This paper describes Monitorix, a video-based traffic surveillance multi-agent system. Monitorix agents are grouped in four tiers, according to the kind of information processing they perform: the sensors and effectors tier, the objective description tier, the application assistant tier, and the user assistant tier. The video analysis algorithms use an adaptive, data-driven, application independent approach to extract features from the video raw data. In spite of the diversity of agent tasks, adaptive learning algorithms are used in most cases. The integration of video analysis algorithms and agent technology is made via a special middle agent called Proxy. Monitorix is a fully decentralised multi-agent system living in a FIPA Platform and using FIPA Agent Communication Language. Tracking of vehicles across nonoverlapping cameras is performed by the Tracker agent, using a traffic model and learning algorithms that tune the model parameters.

Abstract: Multi agent systems comprise multiple, deliberative agents embedded in and recreating patterns of interactions. Each agent's execution consumes considerable storage and calculation capacities. For testing multi agent systems, distributed parallel simulation techniques are required that take the dynamic pattern of composition and interaction of multi-agent systems into account. Analyzing the behavior of agents in virtual, dynamic environments necessitates relating the simulation time to the actual execution time of agents. Since the execution time of deliberative components can hardly be foretold, conservative techniques based on lookahead are not applicable. On the other hand, optimistic techniques become very expensive if mobile agents and the creation and deletion of model components are affected by a rollback. The developed simulation layer of JAMES (a Java Based Agent Modeling Environment for Simulation) implements a moderately optimistic strategy which splits simulation and external deliberation into different threads and allows simulation and deliberation to proceed concurrently by utilizing simulation events as synchronization points.