Design paradigms for meta-control in multi-agent systems

2001

Southeastern Symposium on System Theory, 2001

Agent based computing offers the ability to decentralize computing solutions by incorporating autonomy and intelligence into cooperating, distributed applications. It provides an effective medium for expressing solutions to problems that involve interaction with real-world environments and allows modelling of the world state and its dynamics. This model can be then used to determine how candidate actions affect the world and how to choose the best from a set of actions. Most agent paradigms overlook real-time requirements and computing resource constraints. We discuss the application of agent based computing to RoboCup and examine methods to improve it. In particular, we discuss the incorporation of a meta-level reasoning mechanism that handles individual agent organization, plan generation, task allocation, integration and plan execution. We also propose an architecture where a meta-agent is further enhanced by combining it with system-level resource allocation and optimization. The approach adopted by us unifies agent based computing with adaptive resource management for dynamic real-time systems. The goal is to build and implement a distributed, intelligent, agent based system for dynamic real-time applications

Studies in big data, 2015

The work described in this chapter develops a control framework for modelling, analysis and execution of parallel/distributed time-dependent multi-agent systems. The goal is to clearly separate agent behaviours from crosscutting control concerns which in general are orthogonal to a specific application and transparently affect and regulate its evolution. The approach centres on a minimal computational model based on actors with asynchronous message-passing and actions. Actors are the basic building blocks for modelling the business logic of an application. Actions model activities needed by actors, which have a time duration and require specific computing resources to be executed. Action execution can be either preemptable or not preemptable. Actions are the only abstraction units which have to be reified when passing from model analysis to model implementation. Therefore, the use of actions favours model continuity, i.e., a seamless transformation from model analysis by simulation to model implementation and real time execution. Different pluggable control strategies ranging from pure concurrent to time sensitive (real-time or simulated-time) were implemented. Control strategies are compliant with agent mobility and resource availability. For demonstration purposes, the realized control framework was tailored to the JADE distributed agent infrastructure. This chapter first describes the control framework and its prototyping in JADE. Then presents two case studies. The first one is devoted to a thorough assessment of the timing behaviour and performance of a company help desk system. The second one is concerned with the schedulability analysis of a real-time tasking set. Finally, directions of ongoing and future work are drawn in the conclusions.

Lecture Notes in Computer Science, 2008

In the research discussed here, in addition to extracting meta-models from numerous existing Agent architectures and frameworks, we looked at several Task meta-models, with the aim of creating a more comprehensive Agent meta-model with respect to the analysis, design and development of computer games. From the agent-oriented perspective gained by examining the resultant extensive agent meta-model -named ShaMAN -we then revisit the Task Analysis research domain, and consider what benefits Task Analysis and Modelling may draw from the Agent-oriented paradigm.

Autonomous Agents and Multi-Agent …, 2008

2004

Several agent-oriented methodologies have been proposed over the last few years. Unlike the object-oriented domain and unfortunately for designers, most of the time, each methodology has its own purposes and few standardization works have been done yet, limiting the impact of agent design on the industrial world. This paper tries to find a means to unify three existing methodologies -ADELFE, Gaia and PASSI -by studying their meta-models and the concepts related to them. Comparing a certain number of features at the agent or system level (such as the agent structure, its society or organisation, its interactions capacities or how agents may be implemented) has enabled us to draw up a first version of a unified meta-model proposed as a first step towards interoperability between agent-oriented methodologies.

Abstract We consider the problem of determining metacognition strategies for multiagent systems. A single-agent working in isolation can choose its mode of operation solely based on environmental conditions. However, an agent in a team must consider the modes of other agents as the information available to it and thus its decision process depends critically on the information and decision process of others. We present a model and discuss an example that illustrates the complexity of the problem.

2003

Complex agents operating in open environments must make real-time control decisions on scheduling and planning of domain actions. These decisions are made in the context of limited resources and uncertainty about outcomes of actions. The question of how to sequence domain and control actions without consuming too many resources in the process is the meta-level control problem for a resource-bounded rational agent. Our approach is to design and build a meta-level control agent architecture with bounded computational overhead. It supports decisions on when to accept, delay or reject a new task, how much effort to put into scheduling when reasoning about a new task and whether to reschedule when actual execution performance deviates from expected performance. We show that efficient meta-level control leads to significant improvement in performance and provide empirical results based on handgenerated heuristics and reinforcement learning to support our claim.

Applied Sciences

Multi-agent systems (MAS) allow and promote the development of distributed and intelligent applications in complex and dynamic environments. Applications of this kind have a crucial role in our everyday life, as witnessed by the broad range of domains they are deployed to—such as manufacturing, management sciences, e-commerce, biotechnology, etc. Despite heterogeneity, those domains share common requirements such as autonomy, structured interaction, mobility, and openness—which are well suited for MAS. Therein, in fact, goal-oriented processes can enter and leave the system dynamically and interact with each other according to structured protocols. This special issue gathers 17 contributions spanning from agent-based modelling and simulation to applications of MAS in situated and socio-technical systems.

Lecture Notes in Computer Science, 2001

Coordination, which is the process that an agent reasons about its local actions and the (anticipated) actions of others to try to ensure the community acts in a coherent fashion, is an important issue in multi-agent systems. Coordination is a complicated process that typically consists of several operations: exchanging local information; detecting interactions; deciding whether or not to coordinate; proposing, analyzing, refining and forming commitments; sharing results, and so on. We argue that facets of these different operations can be separated and bundled into two different layers.The lowerlayer pertains to feasibility and implementation operations, i.e., the detailed analysis of candidate tasks and actions, the formation of detailed temporal/resource-specific commitments between agents, and the balancing of non-local and local problem solving activities. In contrast, the upper-layer pertains to domain specific coordination tasks such as the formation of high-level goals and objectives for the agent, and decisions about whether or not to coordinate with other agents to achieve particular goals or bring about particular objectives. Detailed domain state is used at this level to make these high-level coordination decisions. In contrast, decisions at the lower-level do not need to reason about this detailed domain state. However, reasoning about detailed models of the performance characteristics of activities, such as their temporal scope, quality, affects of resource usage on performance, is necessary at this level. In this view, the layers are interdependent activities that operate asynchronously.