This paper lays part of the groundwork for a domain theory of negotiation, that is, a way of classifying interactions so that it is clear, given a domain, which negotiation mechanisms and strategies are appropriate. We define State Oriented Domains, a general category of interaction. Necessary and sufficient conditions for cooperation are outlined. We use the notion of worth in an altered definition of utility, thus enabling agreements in a wider class of joint-goal reachable situations. An approach is offered for conflict resolution, and it is shown that even in a conflict situation, partial cooperative steps can be taken by interacting agents (that is, agents in fundamental conflict might still agree to cooperate up to a certain point). A Unified Negotiation Protocol (UNP) is developed that can be used in all types of encounters. It is shown that in certain borderline cooperative situations, a partial cooperative agreement (i.e., one that does not achieve all agents' goals) might be ...

In our research, we explore the role of negotiation for conflict resolution in distributed search among heterogeneous and reusable agents. We present negotiated search, an algorithm that explicitly recognizes and exploits conflict to direct search activity across a set of agents. In negotiated search, loosely coupled agents interleave the tasks of 1) local search for a solution to some subproblem; 2) integration of local subproblem solutions into a shared solution; 3) information exchange to define and refine the shared search space of the agents; and 4) assessment and reassessment of emerging solutions. Negotiated search is applicable to diverse application areas and problem-solving environments. It requires only basic search operators and allows maximum flexibility in the distribution of those operators. These qualities make the algorithm particularly appropriate for the integration of heterogeneous agents into application systems. The algorithm is implemented in a multi-agent framew...

In this paper, we present an adaptive organizational policy for multi-agent systems called trace. trace allows a collection of multi-agent organizations to dynamically allocate tasks and resources between themselves in order to eciently process an incoming stream of task requests. trace is intended to cope with environments in which tasks have time constraints, and environments that are subject to load variations. trace is made up of two key elements: the task allocation protocol (tap) and the resource allocation protocol (rap). The tap allows agents to cooperatively allocate their tasks to other agents with the capability and opportunity to successfully carry them out. As requests arrive arbitrarily, at any instant, some organizations could have surplus resources while others could become overloaded. In order to minimize the number of lost requests caused by an overload, the allocation of resources to organizations is changed dynamically by the resource allocation protocol (rap), which uses ideas from computational market systems to allocate resources (in the form of problem solving agents) to organizations. We begin by formally dening the task allocation problem, and show that it is np-complete, and hence that centralized solutions to the problem are unlikely to be feasible. We then introduce the task and resource allocation protocols, focussing on the way in which resources are allocated by the rap. We then present some experimental results, which show that trace exhibits high performance despite unanticipated changes in the environment.

In multiagent systems, an agent does not usually have complete information about the preferences and decision making processes of other agents. This might prevent the agents from making coordinated choices, purely due to their ignorance of what others want. This paper describes the integration of a learning module into a communication-intensive negotiating agent architecture. The learning module gives the agents the ability to learn about other agents' preferences via past interactions. Over time, the agents can incrementally update their models of other agents' preferences and use them to make better coordinated decisions. Combining both communication and learning, as two complement knowledge acquisition methods, helps to reduce the amount of communication needed on average, and is justified in situations where communication is computationally costly or simply not desirable (e.g. to preserve the individual privacy). Introduction Multiagent systems are networks of loosely-coupled comp...

Negotiated search, a paradigm for cooperative search and conflict resolution among heterogeneous, reusable, expert agents has been implemented in a flexible framework, TEAM. We present negotiated search and discuss the use of customized negotiated-search strategies that take advantage of specific capabilities and relationships that exist in an agent set. Strategies are dynamically selected based on the individual agents' views of the problem-solving situation and on communicated knowledge about the characteristics of agents in the agent set. We present experiments that show strategies can reduce the amount of search required to find mutuallyacceptable solutions and can improve the quality of those solutions. The use of customized negotiatedsearch strategies has far-reaching implications for the design of agents that are intended to be reusable and for the assembly of agents into application systems. 1 Introduction We present negotiated-search, a distributed-search paradigm for coo...

As distributed systems of computers play an increasingly important role in society, it will be necessary to consider ways in which these machines can be made to interact effectively. Especially when the interacting machines have been independently designed, it is essential that the interaction environment be conducive to the aims of their designers. These designers might, for example, wish their machines to behave efficiently, and with a minimum of overhead required by the coordination mechanism itself. The rules of interaction should satisfy these needs, and others. Formal tools and analysis can help in the appropriate design of these rules.

The natural and social sciences have recognized for some time that when rational agents use a shared resource (a "commons") near its capacity, the resource is doomed. This is the tragedy of the commons. The same problem will also arise when agents in a distributed AI system share a resource. In this paper, we examine the tragedy of the commons as it applies to DAI and discuss several potential solutions, including conventions, privatization of the resource, and mutual coercion mutually agreed upon. Solutions to this problem in DAI testbeds will likely also contribute to understanding how to avoid the tragedy of the commons in many areas of current interest to society, such as fisheries management, preservation of whale and other species populations, global climate change, and preservation of tropical rainforests. The author is also affiliated with the UNH Marine Systems Engineering Laboratory. This paper was submitted in January, 1993, to the 12th International Workshop on Distribute...

Research in distributed AI has led to computational techniques for providing AI systems with rudimentary social skills. This paper gives a brief survey of distributed AI, describing the work that strives for social skills that a person might acquire in kindergarten, and highlighting important unresolved problems facing the field. Introduction In the pursuit of artificial intelligence (AI), it has become increasingly clear that intelligence, whatever it is, has a strong social component [ Bobrow, 1991; Gasser, 1991 ] . Tests for intelligence, such as the Turing test, generally rely on having an (assumedly) intelligent agent evaluate the agent in question by interacting with it. For an agent to be intelligent under such criteria, therefore, it has to be able to participate in a society of agents. Distributed AI (DAI) is the subfield of AI that has, for over a decade now, been investigating the knowledge and reasoning techniques that computational agents might need in order to participa...

. Generating teams of robots that are able to perform their tasks over long periods of time requires the robots to be responsive to continual changes in robot team member capabilities and to changes in the state of the environment and mission. In this article, we describe the L-ALLIANCE architecture, which enables teams of heterogeneous robots to dynamically adapt their actions over time. This architecture, which is an extension of our earlier work on ALLIANCE, is a distributed, behavior-based architecture aimed for use in applications consisting of a collection of independent tasks. The key issue addressed in L-ALLIANCE is the determination of which tasks robots should select to perform during their mission, even when multiple robots with heterogeneous, continually changing capabilities are present on the team. In this approach, robots monitor the performance of their teammates performing common tasks, and evaluate their performance based upon the time of task completion. Robots then use this information throughout the lifetime of their mission to automatically update their control parameters. After describing the L-ALLIANCE architecture, we discuss the results of implementing this approach on a physical team of heterogeneous robots performing proof-of-concept box pushing experiments. The results illustrate the ability of L-ALLIANCE to enable lifelong adaptation of heterogeneous robot teams to continuing changes in the robot team member capabilities and in the environment. 1.

This paper presents a notion of trust for use in multi-agent systems. The role trust can play in various forms of interaction is considered. Trust allows interactions between agents where previously there could be none, and allows the trusting 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. An introductory notation to refer to trusting relationships is presented and further work is discussed. MAAMAW'92, S. Martino al Cimino, Italy, 1992 1 1 Introduction Why cooperate? With whom? To what extent? And when? Previous work in Distributed Artificial Intelligence (DAI) has concentrated on the first of these questions, with little or no thought given to the others. In particular,...