Perspective Of Multi-agent Systems Research Articles

A deep reinforcement learning strategy for autonomous robot flocking

<p>Social behaviors in animals such as bees, ants, and birds have shown high levels of intelligence from a multi-agent system perspective. They present viable solutions to real-world problems, particularly in navigating constrained environments with simple robotic platforms. Among these behaviors is swarm flocking, which has been extensively studied for this purpose. Flocking algorithms have been developed from basic behavioral rules, which often require parameter tuning for specific applications. However, the lack of a general formulation for tuning has made these strategies difficult to implement in various real conditions, and even to replicate laboratory behaviors. In this paper, we propose a flocking scheme for small autonomous robots that can self-learn in dynamic environments, derived from a deep reinforcement learning process. Our approach achieves flocking independently of population size and environmental characteristics, with minimal external intervention. Our multi-agent system model considers each agent’s action as a linear function dynamically adjusting the motion according to interactions with other agents and the environment. Our strategy is an important contribution toward real-world flocking implementation. We demonstrate that our approach allows for autonomous flocking in the system without requiring specific parameter tuning, making it ideal for applications where there is a need for simple robotic platforms to navigate in dynamic environments.</p>

A formal semantics for agent distribution and fault tolerance in Jason

This article provides a formal specification of the distribution and fault-tolerance mechanisms of eJason. The eJason programming language is an extension to the agent-oriented programming language Jason that introduces native support for the transparent distribution of agents as well as fault-tolerance mechanisms. This formal semantics is presented from a multiagent system perspective. It unambiguously describes both the possible evolution of the distributed multiagent system over time and the different instruments for fault detection and fault recovery, hence exposing their strengths. This specification may serve as a reference for researchers interested in the inclusion of similar mechanisms in agent-oriented programming languages. The formal semantics has been mechanized through an (open-source) implementation written in Prolog, which implements both the standard Jason operational semantics, along with the new rules for distribution and fault-tolerance introduced in this article.

Toll-based reinforcement learning for efficient equilibria in route choice

Abstract The problem of traffic congestion incurs numerous social and economical repercussions and has thus become a central issue in every major city in the world. For this work we look at the transportation domain from a multiagent system perspective, where every driver can be seen as an autonomous decision-making agent. We explore how learning approaches can help achieve an efficient outcome, even when agents interact in a competitive environment for sharing common resources. To this end, we consider the route choice problem, where self-interested drivers need to independently learn which routes minimise their expected travel costs. Such a selfish behaviour results in the so-called user equilibrium, which is inefficient from the system’s perspective. In order to mitigate the impact of selfishness, we present Toll-based Q-learning (TQ-learning, for short). TQ-learning employs the idea of marginal-cost tolling (MCT), where each driver is charged according to the cost it imposes on others. The use of MCT leads agents to behave in a socially desirable way such that the is attainable. In contrast to previous works, however, our tolling scheme is distributed (i.e., each agent can compute its own toll), is charged a posteriori (i.e., at the end of each trip), and is fairer (i.e., agents pay exactly their marginal costs). Additionally, we provide a general formulation of the toll values for univariate, homogeneous polynomial cost functions. We present a theoretical analysis of TQ-learning, proving that it converges to a system-efficient equilibrium (i.e., an equilibrium aligned to the system optimum) in the limit. Furthermore, we perform an extensive empirical evaluation on realistic road networks to support our theoretical findings, showing that TQ-learning indeed converges to the optimum, which translates into a reduction of the congestion levels by 9.1%, on average.

Modeling and implementing distributed data mining strategies in JaCa-DDM

This work introduces JaCa-DDM, a novel distributed data mining system founded on the agents and artifacts paradigm, conceived to design, implement, deploy, and evaluate learning strategies. Jason rational agents conform to such strategies to cope with distributed computing environments, where CArtAgO artifacts encapsulate learning algorithms, data sources, evaluation tools, and other services implemented in Weka for data mining tasks. The set of strategies presented in this paper aims at encouraging the use of JaCa-DDM to develop new ones, suited to different needs. For this, our system provides tools to evaluate the resulting models in terms of accuracy, number of instances employed to learn, time of convergence, and volume of communications. Although the emphasis in decision trees, JaCa-DDM can be easily extended by adopting new artifacts, e.g., for meta-learning. The main contributions of the paper are as follows: (i) From the multi-agent systems perspective, our approach illustrates how to exploit the so-called “agentification” of Weka for the sake of code reusability, while preserving the benefits of reasoning at the Belief–Desire–Intention level with Jason; (ii) from the data mining perspective, JaCa-DDM is promoted as an extensible tool to define and test distributed strategies; and (iii) a set of strategies including centralizing, meta-learning and Windowing-based approaches, is carefully analyzed to provide comparisons among them.

Network Based Speed Synchronization Control in the Brush DC Motors Via LQR and Multi-agent Consensus Scheme

In the presented paper, an innovative method to address the problem of synchronous speed in the network connected DC motors is presented by utilizing the linear quadratic regulator control along with the consensus algorithm of the leader following multi-agent system (MAS). From MAS perspective, each chopper fed DC motor is considered as single agent connected through fixed and undirected network. In this study, common lyapunov function is used to ensure switch system stability in such a way that if the $$\mathscr{i}$$ th agent is controllable and observable, then all the agents in the multi-motor system network can reach the consensus i.e. synchronous speed. To validate the proposed method, simulation results are presented using MATLAB by considering the application of the load on the motors and without load too.

Formal modeling and control of cyber-physical manufacturing systems

Cyber-physical manufacturing systems are a new paradigm of manufacturing systems that integrate cyber systems and physical systems to aid smart manufacturing. Cyber-physical manufacturing systems can improve agility and responsiveness and guarantee the quality of products to meet the market requirements. Meanwhile, cyber-physical manufacturing systems also become susceptible to cyber-attacks. In order to improve the trustworthiness of cyber-physical manufacturing systems in the dynamic modeling phase, a cyber-physical manufacturing system formal model based on object-oriented Petri nets is presented from the perspective of multi-agent systems. Some mathematical methods and supporting tools of Petri nets can be utilized to analyze, verify, and validate cyber-physical manufacturing system formal model. To defense the malicious software spreading in cyber-physical manufacturing systems at run-time, a spreading dynamics model is proposed, and its dynamic behaviors are analyzed. A hybrid bifurcation control me...

Retaining Connectivity In Mobile Communication Mesh Networks

This paper deals with the problem of retaining the connectivity in a Mobile Ad-hoc communication mesh Network (MANET). A multi-agent systems perspective is taken, where primary mobile agents (PAs) can only communicate when the relative distance is less than a “visibility range”. PAs form a network that can become disconnected depending on how they move to achieve their global task (which does not include the connectivity maintenance). To retain connectivity, a number of Relay Agents (RAs), whose motion is governed by a command center (CC), are sent to the field to act as “communication bridges” enforcing the global connectivity of the network containing both the PAs and the RAs. Graph-oriented concepts and analysis tools, particularly the minimal spanning tree (MST) notion, are adopted in the present work to analyze the connectivity properties of the network and to establish in real time how many additional RAs are required and how they should move in order to prevent the connectivity loss. Artificial potential fields and finite-time control techniques are utilized to drive the relay agents to their waypoints while avoiding the collisions. Numerical examples confirm the efficacy of the proposed multi-layer control strategy.

Object-oriented Petri nets and π-calculus-based modeling and analysis of reconfigurable manufacturing systems

Reconfigurable manufacturing systems can change the structure of the systems to cope with manufacturing market requirements. Reconfigurability brings about new challenges for reconfigurable manufacturing systems’ development. In order to describe, analyze, and verify the reconfiguration of reconfigurable manufacturing systems, a reconfigurable manufacturing system formal model is proposed from the perspective of multi-agent systems, in which two complementary formalisms, namely, object-oriented Petri nets and π-calculus, are employed as formalisms. The object-oriented Petri nets are utilized to model the initial structure as well as system behaviors of reconfigurable manufacturing systems, while π-calculus is adopted to describe the reconfiguration of reconfigurable manufacturing systems. Some supporting tools of Petri nets and π-calculus can be used to analyze, verify, and validate the reconfigurable manufacturing system formal model. The reconfigurability mechanism and consistency of reconfigurable manufacturing systems can also be analyzed by π-calculus.

Trust and Reputation Models for Multiagent Systems

Finding reliable partners to interact with in open environments is a challenging task for software agents, and trust and reputation mechanisms are used to handle this issue. From this viewpoint, we can observe the growing body of research on this subject, which indicates that these mechanisms can be considered key elements to design multiagent systems (MASs). Based on that, this article presents an extensive but not exhaustive review about the most significant trust and reputation models published over the past two decades, and hundreds of models were analyzed using two perspectives. The first one is a combination of trust dimensions and principles proposed by some relevant authors in the field, and the models are discussed using an MAS perspective. The second one is the discussion of these dimensions taking into account some types of interaction found in MASs, such as coalition, argumentation, negotiation, and recommendation. By these analyses, we aim to find significant relations between trust dimensions and types of interaction so it would be possible to construct MASs using the most relevant dimensions according to the types of interaction, which may help developers in the design of MASs.

Using a multiagent organizational approach to improve knowledge provision for shared decision making in patient-physician relationships: an example from Germany

An integral aspect of sound patient centric healthcare delivery is the notion of shared decision making between patient and provider. This paradigm is well documented in the extant literature. However, what appears to be a stumbling block to date is successfully implementing such an approach within established healthcare processes. This void is causing problems in healthcare delivery in particular in EU countries such as Germany because knowledge sources are distributed and underlie strict privacy policies; while the lack of adequate shared decision making compromises the quality of healthcare delivery and can lead to errors and inefficient workflow. This paper which is based on an extensive research project serves to analyze the provision of personal guidance services for shared decision making in eHealth service networks. By doing so, we address the problem of distributed and privacy-aware knowledge sharing by the formation of agent-based organizations to represent the relationships of patients and physicians and study this problem from the perspective of multiagent systems; i.e., we develop technology enabled collaboration solutions. The efficacy of the proffered decision support system will be demonstrated by a scenario-based evaluation. We contend that such an approach will address the current void.

Reviewing Microgrids from a Multi-Agent Systems Perspective

The construction of Smart Grids leads to the main question of what kind of intelligence such grids require and how to build it. Some authors choose an agent based solution to realize this intelligence. However, there may be some misunderstandings in the way this technology is being applied. This paper exposes some considerations of this subject, focusing on the Microgrid level, and shows a practical example through INGENIAS methodology, which is a methodology for the development of Agent Oriented systems that applies Model Driven Development techniques to produce fully functional Multi-Agent Systems.

Engineering manufacturing control systems using PROSA and delegate MAS

This paper presents a systematic description of a reusable software architecture for multiagent systems in the domain of manufacturing control. The architectural description consolidates the authors' expertise in this area. Until now, the research has taken a manufacturing control perspective of multiagent systems. The research team has focused on providing benefits to the manufacturing control domain by designing a novel type of control system. This paper takes a software architectural perspective of multiagent manufacturing control. The systematic description specifies a software product line architecture for manufacturing control. The paper describes the assets of the software product line architecture and how these assets can be combined.

Multi-agent Negotiation Mechanisms for Statistical Target Classification in Wireless Multimedia Sensor Networks.

The recent availability of low cost and miniaturized hardware has allowedwireless sensor networks (WSNs) to retrieve audio and video data in real worldapplications, which has fostered the development of wireless multimedia sensor networks(WMSNs). Resource constraints and challenging multimedia data volume makedevelopment of efficient algorithms to perform in-network processing of multimediacontents imperative. This paper proposes solving problems in the domain of WMSNs fromthe perspective of multi-agent systems. The multi-agent framework enables flexible networkconfiguration and efficient collaborative in-network processing. The focus is placed ontarget classification in WMSNs where audio information is retrieved by microphones. Todeal with the uncertainties related to audio information retrieval, the statistical approachesof power spectral density estimates, principal component analysis and Gaussian processclassification are employed. A multi-agent negotiation mechanism is specially developed toefficiently utilize limited resources and simultaneously enhance classification accuracy andreliability. The negotiation is composed of two phases, where an auction based approach isfirst exploited to allocate the classification task among the agents and then individual agentdecisions are combined by the committee decision mechanism. Simulation experiments withreal world data are conducted and the results show that the proposed statistical approachesand negotiation mechanism not only reduce memory and computation requi.

MULTI AGENT INTELLIGENT CONTROL OF CENTRALIZED HVAC SYSTEMS

The perspective of Multi Agent Systems (MAS) implementation in thermal control of centralized HVAC systems is considered. MAS are implemented in all functional levels of hierarchical HVAC control system: estimation of representative control variables by agent based softsensing, evaluation of Predicted Mean Vote (PMV) index using intelligent agents, local control applying agent-based Ant Colony Optimization (ACO). A non-square control problem of realization comfort is solved via ACO. Software agents are realized into the framework of JADE – plathform in JAVA. Simulation results are presented for single input-two output (SITO) control system.

Coordination of multiagent systems through explicit valuation of action

The effectiveness of a system relates to its ability to fulfill its functional requirements. When a system consists of a set of collaborating agents, the overall performance may depend more on their ability to work together than on optimizing individual behavior. In fact, the goal of an agent is often at odds with the interest of the collective. The limitations of centralized control for complex systems suggest the need for decentralized coordination. This can be achieved by having each agent take explicit account of the cost of engaging in an activity as well as a measure of reward for effective behavior. The consideration of payoffs and penalties leads to an economic perspective of multiagent systems. In consequence, it is possible to draw on previous work in diverse fields, ranging from game theory to welfare economics and social policy. The promise and limitations of the explicit valuation approach are examined. To illustrate, the behavior of collaborative systems can be interpreted in terms of games of strategy; this approach permits a better understanding of the conditions for globally optimal behavior, as well as strategies for their attainment. The notions of agents and explicit valuation of action are versatile concepts. One indication of the versatility lies in the fact that these concepts cover as a special case the idea of genetic algorithms as a mechanism for learning systems.