. A critical challenge to creating effective multi-agent systems is allowing them to operate effectively in environments where failures (`exceptions') can occur. This paper describes the motivation, progress and plans for work being pursued in this area by the MIT Adaptive Systems and Evolutionary Software research group (http://ccs.mit.edu/ases/). 1. The Challenge: Enabling Robust Open Multi-Agent Systems "open systems ... represent arguably the most important application for multiagent systems" (Wooldridge, Jennings et al. 1999) This paper addresses one simple question: how can we develop effective multi-agent systems out of the diverse and unreliable (buggy, malicious, or simply "dumb") agents and infrastructures we can expect to encounter in open system contexts? This is becoming an increasingly critical question because of emerging changes in the way human organizations work. Globalization, enabled by ubiquitous telecommunications, has increasingly required that organizations ...

Recent years are seeing an increasing need for on-line monitoring of deployed distributed teams of cooperating agents, e.g., for visualization, or performance tracking. However, in deployed systems, we often cannot rely on the agents to communicate their state to the monitoring system: (a) we rarely can change the behavior of already-deployed agents to communicate the required information (e.g., in legacy or proprietary systems); (b) different monitoring goals require different information to be communicated (e.g., agents ’ beliefs vs. plans); and (c) communications may be expensive, unreliable, or insecure. This paper presents a non-intrusive approach based on plan-recognition, in which the monitored agents’ state is inferred from observations of their routine actions. In particular, we focus on inference of the team state based on its observed routine communications, exchanged as part of coordinated task execution. The paper includes the following key novel contributions: (i) a linear time probabilistic plan-recognition algorithm, well-suited for processing communications as observations; (ii) an approach to exploiting general knowledge of teamwork to predict agent responses during normal execution, to reduce monitoring uncertainty; and (iii) a monitoring algorithm that trades expressivity for scalability, representing only certain useful monitoring hypotheses, but allowing for any number of agents and their different activities, to be represented in a single coherent entity. Our empirical evaluation illustrates that monitoring based on observed routine communications enables significant monitoring accuracy, while not being intrusive. The results also demonstrate a key lesson: A combination of complementary low-quality techniques is cheaper, and better, than a single, highly-optimized monitoring approach. 1.

. A critical challenge to creating effective open multi-agent systems is allowing them to operate effectively in the face of potential failures. In this paper we present an experimental evaluation of a set of domain-independent services designed to handle the failure modes ("exceptions") that can occur in such environments, applied to the wellknown "Contract Net" multi-agent system coordination protocol. We show that these services can produce substantially more effective fault handling behavior than standard existing techniques, while allowing simpler agent implementations. Keywords: contract net, failure detection, failure resolution, fault handling, multiagent 1 1. The Challenge A critical challenge to creating effective agent-based systems is making them robust in the face of potential failures. Most work to date on multi-agent systems has focused, however, on supporting such basic functionality such as matchmaking (Decker, Sycara et al. 1997) and inter-agent communication (Fin...

Teamwork demands agreement among teammembers to collaborate and coordinate effectively. When a disagreement between teammates occurs (due to failures), team-members should ideally diagnose its causes, to resolve the disagreement. Such diagnosis of social failures can be expensive in communication and computation overhead, which previous work did not address. We present a novel design space of diagnosis algorithms, distinguishing several phases in the diagnosis process, and providing alternative algorithms for each phase. We then combine these algorithms in different ways to empirically explore specific design choices in a complex domain, on thousands of failure cases. The results show that centralizing the diagnosis disambiguation process is a key factor in reducing communications, while run-time is affected mainly by the amount of reasoning about other agents. These results contrast sharply with previous work in disagreement detection, in which distributed algorithms reduce communications.

Agents in dynamic multi-agent environments must monitor their peers to execute individual and group plans. A key open question is how much monitoring of other agents' states is required to be effective: The Monitoring Selectivity Problem. We investigate this question in the context of detecting failures in teams of cooperating agents, via Socially-Attentive Monitoring, which focuses on monitoring for failures in the social relationships between the agents. We empirically and analytically explore a family of socially-attentive teamwork monitoring algorithms in two dynamic, complex, multi-agent domains, under varying conditions of task distribution and uncertainty. We show that a centralized scheme using a complex algorithm trades correctness for completeness and requires monitoring all teammates. In contrast, a simple distributed teamwork monitoring algorithm results in correct and complete detection of teamwork failures, despite relying on limited, uncertain knowledge, and monitoring only key agents in a team. In addition, we report on the design of a socially-attentive monitoring system and demonstrate its generality in monitoring several coordination relationships, diagnosing detected failures, and both on-line and off-line applications.

A Multi-Agent System (MAS) is an organization of coordinated autonomous agents that interact in order to achieve particular, possible common goals. Considering real world organizations as an analogy, this paper proposes architectural styles for MAS which adopt concepts from organizational theories. The styles are modeled using the i* framework which o#ers the notions of actor, goal and actor dependency and specified in Formal Tropos. They are evaluated with respect to a set of software quality attributes, such as predictability or adaptability. In addition, we conduct a comparative study of organizational and conventional software architectures using the mobile robot control example from the Software Engineering literature. The research is conducted in the context of Tropos, a comprehensive software system development methodology.

The first wave of agent implementation toolkits focussed mostly on providing APIs for agent communication. We believe that new toolkits should focus on the public dissemination of complete agent architectures that provide significant value over building software agents from scratch. DECAF (Distributed, Environment Centered Agent Framework) is a software toolkit for the rapid design, development, and execution of “intelligent ” agents to achieve solutions in complex software systems. DECAF is based on the premise that execution of the actions required to accomplish a task specified by an agent program is similar to a traditional operating system executing a sequence of user requests. In the same fashion that an operating system provides an environment for the execution of a user request, an agent framework provides the needed environment for the execution of agent actions. The agent environment includes the ability to communicate with other agents, efficiently maintain the current state of an executing agent, and select an execution path from a set of possible execution paths so as to support persistent, flexible, and robust actions. From a research community perspective, DECAF provides a modular platform for evaluating and disseminating results in agent architectures, including communication, planning,

Software agent marketplaces require the development of new architectures, which are capable of coping with unreliable computational and network infrastructures, limited trust among independently developed agents and the possibility of systemic failures. In analogy with human societies, agent marketplaces will benefit from the introduction of appropriate electronic exception handling institutions, whose role will be to help guarantee efficiency and fairness in the face of these challenges. This paper presents a research methodology for designing and evaluating such electronic institutions. It also describes how the methodology has been applied in order to design and evaluate an exception handling architecture for robust software agent marketplaces based on the contract net protocol.

Abstract. Exploring agent conversation in the context of fine-grained agent coordination research has raised several intellectual questions. The major issues pertain to interactions between different agent conversations, the representations chosen for different classes of conversations, the explicit modeling of interactions between the conversations, and how to address these interactions. This paper is not so ambitious as to attempt to address these questions, only frame them in the context of quantified, scheduling-centric multi-agent coordination research. 1

Abstract not found