Showing papers on "Multi-agent system published in 1992"
01 Jan 1992
TL;DR: This dissertation has focused on implementing a platform for reasoning about emotions that supports the testing of psychological theory and model a multi-agent world and gives simple affective life to agents in the form of rudimentary emotions and emotion-induced actions.
Abstract: The problem we have addressed in this dissertation is that of designing a pragmatic and rich computer representation of emotions that is at least congenial with psychological theory. Our solution has focused on implementing a platform for reasoning about emotions that supports the testing of such theories.
In the platform we model a multi-agent world and give simple affective life to agents in the form of rudimentary emotions and emotion-induced actions. In addition the agents are able to reason about emotion episodes that take place in one another's lives. The implementation includes representations for twenty-four emotion types (based on the work Ortony et al., 1988) and 1400 emotion-induced actions. Agents have rudimentary personalities, including an interpretive component which causes them to construe the world in idiosyncratic ways leading to emotional states, and an expressive component which give agents a unique profile for manifesting their emotions. Agents keep internal models of the concerns of other agents which allow them to explain the emotional episodes of other agents by classifying them as instances in which one or more of the twenty-four emotion types arise.
The implementation is simulation-based. It has been run with up to forty agents at a time. Situations arise in the modeled world, agents respond to some of these in their own unique, emotional, ways. Emotion-induced actions are generated which may be placed back in the simulation queue and further perturb the system. Other agents observe and explain the situations using both strong-theory reasoning based on a set of emotion rules, and weak-theory reasoning using a case-based heuristic classification system.
546 citations
01 Jan 1992
TL;DR: This thesis defines a model of agents and multi-agent systems, and then defines two execution models, which describe how agents may act and interact, and a number of logics, with various properties, are developed in this way.
Abstract: THE aim of this thesis is to investigate logical formalisms for describing, reasoning about, specifying, and perhaps ultimately verifying the properties of systems composed of multiple intelligent computational agents. There are two obvious resources available for this task. The first is the (largely AI) tradition of reasoning about the intentional notions (belief, desire, etc.). The second is the (mainstream computer science) tradition of temporal logics for reasoning about reactive systems. Unfortunately, neither resource is ideally suited to the task: most intentional logics have little to say on the subject of agent architecture, and tend to assume that agents are perfect reasoners, whereas models of concurrent systems from mainstream computer science typically deal with the execution of individual program instructions. This thesis proposes a solution which draws upon both resources. It defines a model of agents and multi-agent systems, and then defines two execution models, which describe how agents may act and interact. The execution models define what constitutes an acceptable run of a system. A run may then act as a model for a temporal logic; this logic can subsequently be used to describe and reason about multi-agent systems. A number of logics, with various properties, are developed in this way. Several detailed examples are presented, showing how the logics may be used for specifying and reasoning about multi-agent systems. The thesis includes a detailed literature survey.
152 citations
29 Jul 1992
TL;DR: A discussion of trust is presented which focuses on multiagent systems, from the point of view of one agent in a system, with the view that trust allows interactions between agents where there may have been no effective interaction possible before trust.
Abstract: A discussion of trust is presented which focuses on multiagent systems, from the point of view of one agent in a system. The roles trust plays in various forms of interaction are considered, with the view that trust allows interactions between agents where there may have been no effective interaction possible before trust. Trust allows parties to acknowledge that, whilst there is a risk in relationships with potentially malevolent agents, some form of interaction may produce benefits, where no interaction at all may not. In addition, accepting the risk allows the trusting agent to prepare itself for possibly irresponsible or untrustworthy behaviour, thus minimizing the potential damage caused. A formalism is introduced to clarify these notions, and to permit computer simulations. An important contribution of this work is that the formalism is not allen-compassing: there are some notions of trust that are excluded. What it describes is a specific view of trust.
89 citations
Proceedings Article•
01 Jan 1992
78 citations
Book•
12 Nov 1992
70 citations
TL;DR: A generic model is described, the Recursive Negotiation Model (RNM), that can serve as a basis for classifying and specifying where conflict resolution among multiple experts, viewpoints, or types of reasoning is needed in building a sophisticated CDPS system.
Abstract: Research in Cooperative Distributed Problem Solving (CDPS) considers how problem-solving tasks should be allocated among a group of agents and how the agents should coordinate their actions to achieve effective problem solving. For some CDPS systems, negotiation plays an important role in how agents cooperate. We define negotiation as the process of information exchange by which the agents act to resolve inconsistent views and to reach agreement on how they should work together in order to cooperate effectively. We describe a generic model, the Recursive Negotiation Model (RNM), that can serve as a basis for classifying and specifying where conflict resolution among multiple experts, viewpoints, or types of reasoning is needed in building a sophisticated CDPS system. This model defines where and how negotiation can be applied during problem solving based on structuring problem solving into four stages: problem formulation, focus-of-attention, allocation of goals or tasks to agents, and achievement of goals or tasks. We further discuss how the degree of agent participation in control decisions, including decisions about assigning responsibility to agents, influences the nature of negotiation within a particular system. Through this model, we emphasize that negotiation may be a recursive, complex, and pervasive process that is used to resolve conflicts in both domain-level and control-level problem solving. Finally, we survey existing negotiation frameworks and how they relate to our generic model.
63 citations
TL;DR: GRATE offers a paradigm shift for the construction of multi-agent systems in which the role of configuring preexisting knowledge becomes an integral component and the designer can utilise the inbuilt knowledge and augment it, if necessary, with domain specific information.
Abstract: As the deployment of expert systems has spread into more complex and sophisticated environments, so inherent technological limitations have been observed. As a technique for overcoming this complexity barrier, researchers have started to build systems composed of multiple, cooperating components. These systems tend to fall into two distinct categories: systems which solve a particular problem, such as speech recognition or vehicle monitoring, and systems which are general to some extent. GRATE is a general framework which enables an application builder to construct multi-agent systems for the domain of industrial process control. Unlike other cooperation frameworks, GRATE embodies a significant amount of inbuilt knowledge related to cooperation and control which can be utilised during system building. This approach offers a paradigm shift for the construction of multi-agent systems in which the role of configuring preexisting knowledge becomes an integral component. Rather than starting from scratch the designer can utilise the inbuilt knowledge and augment it, if necessary, with domain specific information. The GRATE architecture has a clear separation of concerns and has been applied to real-world problems in the domains of electricity transportation management and diagnosis of a particle accelerator beam controller.
63 citations
01 May 1992
TL;DR: It is argued that organizational changes will have to be made before the second set of forces can be applied to effectively counter the first.
Abstract: The real-time operations of electric power networks are subject to two sets of forces. The first, including deregulation movements and growing environmental concerns, is acting to increase the complexity of operations. The second, including new computer technologies and emerging knowledge-based agents, provides some means for handling additional complexity. The authors argue that organizational changes will have to be made before the second set of forces can be applied to effectively counter the first. To make this argument, a framework for discussing organizational structures is presented. Then the structures of the two generations of computer-based, multiagent systems that have been developed for operations are reviewed. It is pointed out that these structures are well-suited to the algorithmic tasks involved in operations but not to the knowledge-based tasks. The authors conclude with some suggestions for research into alternative structures. >
37 citations
Proceedings Article•
30 Aug 1992TL;DR: A sound proof system for the logic is presented, and some short examples are given, showing how the logic might be used to specify desirable properties of multi-agent systems.
Abstract: This paper presents a first-order branching time temporal logic that is suitable for describing and reasoning about a wide class of computational multi-agent systems The logic is novel in that it supports reasoning about the beliefs, actions, goals, abilities and structure of groups of agents A sound proof system for the logic is presented, and some short examples are given, showing how the logic might be used to specify desirable properties of multi-agent systems
30 citations
TL;DR: This paper describes the experimental testbed that is currently developing to facilitate the investigation of various constraint-based strategies for addressing the coordination issues inherent in cooperative distributed problem-solving domains.
Abstract: Organizations influence many aspects of our lives. They exist for one reason: they can accomplish things that individuals cannot. While recent work in high-autonomy systems has shown that autonomy is a critical issue in artificial intelligence (AI) systems, these systems must also be able to cooperate with and rely on one another to deal with complex problems. The autonomy of such systems must be flexible, in order that agents may solve problems on their own as well as in groups. We have developed a model of distributed problem solving in which coordination of problem-solving agents is viewed as a multiagent constraint-satisfaction planning problem. This paper describes the experimental testbed that we are currently developing to facilitate the investigation of various constraint-based strategies for addressing the coordination issues inherent in cooperative distributed problem-solving domains.
23 citations
TL;DR: In this article, the authors report the implementation and operational experience of a knowledge-based system for voltage/var control in a real-time environment, which is a multi-functional, multi-paradigm tool fully embedded and consistent with the existing decision support system.
TL;DR: This paper shows how such an architecture is being applied by a Dutch consortium to a part of the Columbus User Support Organisation (USO) and shows how MAS techniques may be applied to the In Orbit Infrastructure Ground Segment (IOI-GS) and, in particular, to the Columbus USO.
01 Apr 1992
TL;DR: A theory of the interaction among agents and a formal semantics for their interactions are presented, which provides some insights into designing distributed intelligence systems, but also helps in their validation.
Abstract: A theory of the interaction among agents and a formal semantics for their interactions are presented. The semantics of the messages exchanged, not the process of exchanging them, is emphasized. A recent theory of communication that gives the object model-theoretic semantics for speech acts is applied to this problem. This allows the important properties of protocols to be formalized abstractly, i.e. at the level of the application, not of the implementation. Further constraints on good designs can also be stated, which simplify the requirements imposed on the member agents. The resulting theory not only provides some insights into designing distributed intelligence systems, but also helps in their validation. As an example, it is applied to a logical reconstruction of the classical Contract Net protocol. >
23 Aug 1992
TL;DR: A new architecture which based on distributed artificial intelligence techniques treats the inherent problems of natural language processing and contains linguistic laws which allow the management of the communication between agents, without appealing to a central control.
Abstract: Natural language processing raises the problem of ambiguities and multiple solutions which follow from them. The knowledge gained when using the morphosyntactic analyser CRISSTAL showed how necessary it was to overcome this issue. The architecture with sequential levels, in which each module corresponds to a linguistic level (pretreatments, morphology, syntax, semantics) has shown its limits. A sequential architecture does not allow a real exchange between different modules. This leads to the non availability of the linguistic information for the reduction of ambiguities, at the moment they are needed.The necessity for cooperation between different modules has lead us to envisage a new architecture which stems from the techniques of distributed artificial intelligence.The paper presents this new architecture which based on distributed artificial intelligence techniques treats the inherent problems of natural language processing. One of the originalities resides in the distributed treatment of sentence analysis (as apposed to a classic sequential treatment) and in the introduction of linguistic laws which allow the management of the communication between agents, without appealing to a central control. The Talisman system is an environment which integrates linguistic tools where different agents can bring into use different methods such as symbolic and/or statistic ones.The Talisman system contributes to the following points in the linguistic domain:•the restriction of ambiguities by agent cooperation,•rendering structures less complex by using local grammar rules,•the treatment of uncertain information.It can:•function with partial analyses at different classical levels of analysis,•change strategies according to the applications or the corpus used,•use linguistic laws which are easily modifiable.At the implementation level, the system brings openess to the modification of dictionaries, grammars and strategies of analysis, and the necessary mechanisms for the integration of new modules.Talisman is a linguistic environment based on the most recent techniques used in software engineering environments. It provides mechanisms for data and control integration of linguistic tools.This paper is organized as follows. After a short overview in section 2 about the problems of sequential architectures, we establish the main objectives of our work in section 3. We present, in section 4, the contribution of multi-agents systems "governed by laws". In section 5, we define the structure of an agent and of its society. The implementation is presented in section 6.
01 Jul 1992
TL;DR: A bulletin board model is proposed as a flexible paradigm for supporting the coexistence of disparate agents and for investigating agent interaction mechanisms.
Abstract: Agent dialogue and interaction protocols, which are important in achieving intelligent collective behavior from a set of heterogeneous agents in real-time computer systems, are addressed. The characteristics of such agents are analyzed and classified into several generic classes. A bulletin board model is proposed as a flexible paradigm for supporting the coexistence of disparate agents and for investigating agent interaction mechanisms. A set of protocols for agent dialogue and interaction is suggested. The analysis places a strong emphasis on the time in properties of such protocols and on the need to satisfy temporal constraints in real-time environments. >
TL;DR: The method advocated is to design the individual agents so that they maintain symbolic representations of emergent behaviour which can then be used as a basis for building higher level behaviours.
Abstract: This paper addresses the question of how to exploit emergent behaviour in the design of multi-agent systems. The method advocated is to design the individual agents so that they maintain symbolic representations of emergent behaviour which can then be used as a basis for building higher level behaviours. The paper falls into four main parts: a description of behaviour-based agents, a discussion of emergent behaviour, a description of a methodology for developing agents which exploit emergent behaviour, and a practical example of the application of these ideas to the development of a simulated multi-agent system.
09 Aug 1992
TL;DR: Based on the features of both distributed artificial intelligence and load forecasting techniques, a distributed problem-solving system for short-term load forecasting is proposed, a multi-agent system that can compute autonomously and cooperate with other agents to reason an accurate and satisfactory solution for load forecasting.
Abstract: Based on the features of both distributed artificial intelligence and load forecasting techniques, a distributed problem-solving system for short-term load forecasting is proposed. Such a distributed paradigm is a multi-agent system, each processing agent of which can compute autonomously and cooperate with other agents to reason an accurate and satisfactory solution for load forecasting. The designed load forecasting system solves problems using three basic moduli: blackboard module, knowledge sources, and control mechanism. In addition, to achieve a high degree of accuracy in load forecasting, the techniques are embedded in the domain knowledge source. The system was implemented by an expert system and tested. It was found that the developed distributed system is a valuable tool to system operators for short-term load forecasting. >
TL;DR: It was never possible to define permanently the operator's role when supported by an intelligent machine, so now are the authors are seeing the emergence of negotiation and dialogue between operator and machine for dynamic allocation of control tasks?
Abstract: Artificial Intelligence has been applied in process control, using a range of AI techniques, such as rules, fuzzy logic and neural nets, and architectures, including blackboards, layered architectures and, most recently, distributed and multi-agent architectures. In the early days, expert systems in process control were seen as a way of possibly removing the operator from the control loop. But as applications have become more complex and the processes are becoming managed rather than merely controlled, the role of the operator has changed. It was never possible to define permanently the operator's role when supported by an intelligent machine, so now are we are seeing the emergence of negotiation and dialogue between operator and machine for dynamic allocation of control tasks?
28 May 1992
TL;DR: The paper studies the integration of communication in independent expert system shell environments through a client-server architecture.
Abstract: Distributed artificial intelligence is concerned with coordination strategies of decentralized groups of intelligent computational agents. The paper discusses multi-agent systems and the coordination of intelligent behavior among a collection of such agents. A prerequisite for the communication among independent expert systems is the existence of suitable tools. The paper studies the integration of communication in independent expert system shell environments through a client-server architecture. >
01 Jan 1992
TL;DR: Maite is a knowledge-based operator assistance system for trouble-shooting a large telecommunications network capable of responding in real time to multiple alarms coming from different digital switches in the network.
Abstract: MAITE is a knowledge-based operator assistance system for trouble-shooting a large telecommunications network Maite is capable of responding in real time to multiple alarms coming from different digital switches in the network It provides the telecommunications operator with advice and guidance for diagnosis and repair tasks Maite is based on a multi-agent blackboard architecture Agents embody general knowledge of diagnosis and repair or specific expertise in the various components of the network They can work in parallel interpreting alarms and performing diagnostic and repair tasks Agents’ reasoning methods include temporal, model based, and expectation-based reasoning Coordination is assured by organisation and agent specialization Overall control of the system is achieved by a control agent that supervises communication with external units and integrates partial results from specialist agents It is robust, can provide detailed explanations of its actions, and is easily extensible
18 Oct 1992
TL;DR: A remote procedure dispatching (RPD) scheme and an ambassador model provides a supplement to the client-server model for organizational modeling and lies in its flexibility for agent-oriented unilateral protocol selection and protocol change.
Abstract: The concepts of virtual agents and virtual communities are introduced. Based on these concepts, a remote procedure dispatching (RPD) scheme and an ambassador model are proposed. While virtual agents and virtual communities are images of remote agents dynamically created and maintained by a local agent as computer network domain objects, the RPD scheme supplements remote procedure call by allowing remote procedures to be dispatched to remote sites, and the ambassador model allows an image or an ambassador of an agent to be dispatched to remote hosts. Then, the virtual agents can be organized into virtual communities for coordination purposes. The major advantages of this approach lies in its flexibility for agent-oriented unilateral protocol selection and protocol change. Thus, it provides a supplement to the client-server model for organizational modeling. A prototype design and implementation are presented. The basic ideas are illustrated with an example in information retrieval and business administration. >
01 Jan 1992
TL;DR: The distributed model that is developed in this dissertation has been utilized for the development of a task-executor that translates a robot-independent plan into robot actions and allows for the incorporation of quantitative measures for performance evaluation.
Abstract: In this dissertation, we develop a distributed model of robot task synchronization and interaction by utilizing developments from universal algebra, the theory of algebraic specifications, and Petri net theory. The novelty of this model is that it allows for the dynamic specification of the distributed semantics of multiagent systems, in general, and robot systems, in particular. This is in contrast to the traditional practice of the static specification of the interleaving semantics of such systems.
More specifically, the distributed model that is developed in this dissertation provides the means to perform the following: (1) The specification of a plan for the execution of a task. (2) The hierarchical decomposition of a plan. (3) The distribution and localization of the resulting plan components to the appropriate agents. (4) The determination and specification of the temporal order on the control flow of tasks of multiple agents. (5) The synchronization of concurrent tasks when necessary. (6) The determination and specification of the interactions of concurrent tasks.
Thus, the distributed model that is developed in this dissertation is a significant step towards the realization of robot systems that exhibit a high degree of autonomy. The realization of such systems is desirable because (i) they will greatly enhance the task-executing versatility of contemporary robot systems, and (ii) they will allow for robotic applications in unstructured environments such as space and underwater exploration, contaminated areas, and agriculture.
The distributed model that we develop in this dissertation has been utilized for the development of a task-executor that translates a robot-independent plan into robot actions. Qualitative results from two case studies illustrate the plausibility of the developed model. The model also allows for the incorporation of quantitative measures for performance evaluation. However, in the absence of a benchmark for comparison, from any prior work that we are aware of, quantitative results have not been obtained.
The distributed model that is developed in this dissertation is also applicable to domains other than the intelligent robot control. Intuitively, it seems most appropriate in dynamic domains which are characterized by an inherent hierarchical structure.
TL;DR: A methodology that allows collections of intelligent system to automatically learn communication strategies, so that they can exchange information and coordinate their problem solving activity.
Abstract: We present a methodology that allows collections of intelligent system to automatically learn communication strategies, so that they can exchange information and coordinate their problem solving activity. In our methodology communication between agents is determined by the agents themselves, which consider the progress of their individual problem solving activities compared to the communication needs of their surrounding agents. Through learning, communication lines between agents might be established or disconnected, communication frequencies modified, and the system can also react to dynamic changes in the environment that might force agents to cease to exist or to be added. We have established dynamic, quantitative measures of the usefulness of a fact, the cost of a fact, the work load of an agent, and the selfishness of an agent (a measure indicating an agent's preference between transmitting information versus performing individual problem solving), and use these values to adapt the communication between intelligent agents. In this paper we present the theoretical foundations of our work together with experimental results and performance statistics of networks of agents involved in cooperative problem solving activities.© (1992) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.
29 Jul 1992
TL;DR: This paper describes the results of a short programme of experiments investigating the dynamic behaviour of systems of coupled agents, which respond to changes in each other's state.
Abstract: This paper describes the results of a short programme of experiments investigating the dynamic behaviour of systems of coupled agents. The systems investigated are composed of small numbers of simple agents, which respond to changes in each other's state. The mechanism used to determine the behaviour of individual agents is analogous to simple field theories in which the motion of particles is determined by a potential field. The agent potentials can be thought of as representing motivations such as fear, hunger and thirst.