Showing papers on "Multi-agent system published in 2013"
TL;DR: A novel control strategy for multi-agent coordination with event-based broadcasting is presented, in which each agent decides itself when to transmit its current state to its neighbors and the local control laws are based on these sampled state measurements.
1,077 citations
TL;DR: This technical brief considers the distributed consensus problems for multi-agent systems with general linear and Lipschitz nonlinear dynamics and finds that the adaptive consensus protocols here can be implemented by each agent in a fully distributed fashion without using any global information.
Abstract: This technical brief considers the distributed consensus problems for multi-agent systems with general linear and Lipschitz nonlinear dynamics. Distributed relative-state consensus protocols with an adaptive law for adjusting the coupling weights between neighboring agents are designed for both the linear and nonlinear cases, under which consensus is reached for all undirected connected communication graphs. Extensions to the case with a leader-follower communication graph are further studied. In contrast to the existing results in the literature, the adaptive consensus protocols here can be implemented by each agent in a fully distributed fashion without using any global information.
708 citations
TL;DR: A combinational measurement approach to event design and a new iterative event-triggered algorithm where continuous measurement of the neighbor states is avoided are proposed, which reduces the amount of communication and lowers the frequency of controller updates in practice.
689 citations
Book•
31 Dec 2013
TL;DR: Optimal cooperative control is introduced and neural adaptive design techniques for multi-agent nonlinear systems with unknown dynamics, which are rarely treated in literature are developed.
Abstract: Cooperative Control of Multi-Agent Systems extends optimal control and adaptive control design methods to multi-agent systems on communication graphs. It develops Riccati design techniques for general linear dynamics for cooperative state feedback design, cooperative observer design, and cooperative dynamic output feedback design. Both continuous-time and discrete-time dynamical multi-agent systems are treated. Optimal cooperative control is introduced and neural adaptive design techniques for multi-agent nonlinear systems with unknown dynamics, which are rarely treated in literature are developed. Results spanning systems with first-, second- and on up to general high-order nonlinear dynamics are presented. Each control methodology proposed is developed by rigorous proofs. All algorithms are justified by simulation examples. The text is self-contained and will serve as an excellent comprehensive source of information for researchers and graduate students working with multi-agent systems.
542 citations
TL;DR: An event triggering scheme is designed based on a quadratic Lyapunov function that is sampled-data and distributed in the sense that the event detector uses only neighbor information and local computation at discrete sampling instants.
489 citations
TL;DR: It is established that, under the assumptions that each agent is asymptotically null controllable with bounded controls and that the network is connected or jointly connected, semi-global leader-following consensus of the multi-agent system can be achieved.
Abstract: This paper investigates the problem of leader-following consensus of a linear multi-agent system on a switching network. The input of each agent is subject to saturation. Low gain feedback based distributed consensus protocols are developed. It is established that, under the assumptions that each agent is asymptotically null controllable with bounded controls and that the network is connected or jointly connected, semi-global leader-following consensus of the multi-agent system can be achieved. Numerical examples are presented to illustrate this result.
456 citations
TL;DR: This paper presents the development and evaluation of a novel system of multiagent reinforcement learning for integrated network of adaptive traffic signal controllers (MARLIN-ATSC), and shows unprecedented reduction in the average intersection delay.
Abstract: Population is steadily increasing worldwide, resulting in intractable traffic congestion in dense urban areas. Adaptive traffic signal control (ATSC) has shown strong potential to effectively alleviate urban traffic congestion by adjusting signal timing plans in real time in response to traffic fluctuations to achieve desirable objectives (e.g., minimize delay). Efficient and robust ATSC can be designed using a multiagent reinforcement learning (MARL) approach in which each controller (agent) is responsible for the control of traffic lights around a single traffic junction. Applying MARL approaches to the ATSC problem is associated with a few challenges as agents typically react to changes in the environment at the individual level, but the overall behavior of all agents may not be optimal. This paper presents the development and evaluation of a novel system of multiagent reinforcement learning for integrated network of adaptive traffic signal controllers (MARLIN-ATSC). MARLIN-ATSC offers two possible modes: 1) independent mode, where each intersection controller works independently of other agents; and 2) integrated mode, where each controller coordinates signal control actions with neighboring intersections. MARLIN-ATSC is tested on a large-scale simulated network of 59 intersections in the lower downtown core of the City of Toronto, ON, Canada, for the morning rush hour. The results show unprecedented reduction in the average intersection delay ranging from 27% in mode 1 to 39% in mode 2 at the network level and travel-time savings of 15% in mode 1 and 26% in mode 2, along the busiest routes in Downtown Toronto.
437 citations
TL;DR: This study proposes a secondary voltage and frequency control scheme based on the distributed cooperative control of multi-agent systems that is fully distributed such that each distributed generator only requires its own information and the information of its neighbours on the communication digraph.
Abstract: This study proposes a secondary voltage and frequency control scheme based on the distributed cooperative control of multi-agent systems. The proposed secondary control is implemented through a communication network with one-way communication links. The required communication network is modelled by a directed graph (digraph). The proposed secondary control is fully distributed such that each distributed generator only requires its own information and the information of its neighbours on the communication digraph. Thus, the requirements for a central controller and complex communication network are obviated, and the system reliability is improved. The simulation results verify the effectiveness of the proposed secondary control for a microgrid test system.
432 citations
TL;DR: A distributed adaptive law is proposed for each follower based on local information of neighboring agents and the leader if this follower is an informed agent and a distributed leader–follower consensus problem in multi-agent systems with unknown nonlinear dynamics is investigated.
371 citations
TL;DR: This thematic issue reviews progress in spatial agent-based models along the lines of four methodological challenges: design and parameterizing of agent decision models, verification, validation and sensitivity analysis, integration of socio-demographic, ecological, and biophysical models, and spatial representation.
Abstract: Departing from the comprehensive reviews carried out in the field, we identify the key challenges that agent-based methodology faces when modeling coupled socio-ecological systems. Focusing primarily on the papers presented in this thematic issue, we review progress in spatial agent-based models along the lines of four methodological challenges: (1) design and parameterizing of agent decision models, (2) verification, validation and sensitivity analysis, (3) integration of socio-demographic, ecological, and biophysical models, and (4) spatial representation. Based on this we critically reflect on the future work that is required to make agent-based modeling widely accepted as a tool to support the real world policy. Progress of agent-based methodology in modeling coupled socio-ecological systems.Key methodological challenges for ABM.Societal issues and critical reflection on the prospects of ABM.
371 citations
TL;DR: Simulation results clearly indicate that the agent-based management is effective in resource management among multiple microgrids economically and profitably.
Abstract: Microgrid is a combination of distributed generators, storage systems, and controllable loads connected to low-voltage network that can operate either in grid-connected or in island mode. High penetration of power at distribution level creates such multiple microgrids. This paper proposes a two-level architecture for distributed-energy-resource management for multiple microgrids using multiagent systems. In order to match the buyers and sellers in the energy market, symmetrical assignment problem based on naive auction algorithm is used. The developed mechanism allows the pool members such as generation agents, load agents, auction agents, grid agents, and storage agents to participate in market. Three different scenarios are identified based on the supply-demand mismatch among the participating microgrids. At the end of this paper, two case studies are presented with two and four interconnected microgrids participating in the market. Simulation results clearly indicate that the agent-based management is effective in resource management among multiple microgrids economically and profitably.
TL;DR: A network-based consensus control protocol under a directed graph with a new delay-dependent stability criterion for an error system is derived by constructing a novel Lyapunov–Krasovskii functional with digraph information.
TL;DR: In this article, the convergence analysis of a class of distributed constrained non-convex optimization algorithms in multi-agent systems is studied and it is proved that consensus is asymptotically achieved in the network and that the algorithm converges to the set of Karush-Kuhn-Tucker points.
Abstract: We introduce a new framework for the convergence analysis of a class of distributed constrained non-convex optimization algorithms in multi-agent systems. The aim is to search for local minimizers of a non-convex objective function which is supposed to be a sum of local utility functions of the agents. The algorithm under study consists of two steps: a local stochastic gradient descent at each agent and a gossip step that drives the network of agents to a consensus. Under the assumption of decreasing stepsize, it is proved that consensus is asymptotically achieved in the network and that the algorithm converges to the set of Karush-Kuhn-Tucker points. As an important feature, the algorithm does not require the double-stochasticity of the gossip matrices. It is in particular suitable for use in a natural broadcast scenario for which no feedback messages between agents are required. It is proved that our results also holds if the number of communications in the network per unit of time vanishes at moderate speed as time increases, allowing potential savings of the network's energy. Applications to power allocation in wireless ad-hoc networks are discussed. Finally, we provide numerical results which sustain our claims.
TL;DR: A systematic methodology for designing local agent objective functions that guarantees an equivalence between the resulting Nash equilibria and the optimizers of the system level objective and that the resulting game possesses an inherent structure that can be exploited in distributed learning, e.g., potential games.
Abstract: The central goal in multiagent systems is to design local control laws for the individual agents to ensure that the emergent global behavior is desirable with respect to a given system level objective. Ideally, a system designer seeks to satisfy this goal while conditioning each agent's control law on the least amount of information possible. This paper focuses on achieving this goal using the field of game theory. In particular, we derive a systematic methodology for designing local agent objective functions that guarantees (i) an equivalence between the resulting Nash equilibria and the optimizers of the system level objective and (ii) that the resulting game possesses an inherent structure that can be exploited in distributed learning, e.g., potential games. The control design can then be completed utilizing any distributed learning algorithm which guarantees convergence to a Nash equilibrium for the attained game structure. Furthermore, in many settings the resulting controllers will be inherently robust to a host of uncertainties including asynchronous clock rates, delays in information, and component failures.
TL;DR: A new class of observer-based control algorithms are designed for achieving finite-time consensus tracking in multi-agent systems with a single active leader, where each agent can only share its position states with its neighbors.
TL;DR: This paper presents a distributed containment control approach for uncertain nonlinear strict-feedback systems with multiple dynamic leaders under a directed graph topology where the leaders are neighbors of only a subset of the followers.
TL;DR: This paper studies the distributed consensus tracking problem of linear higher-order multi-agent systems with switching directed topologies and occasionally missing control inputs and appropriately constructing a switching Lyapunov function and using tools from the M -matrix theory.
TL;DR: In this article, the authors developed a general purpose (non-disease-specific) computational/artificial intelligence (AI) framework to address the challenges of modern healthcare system, rapidly expanding costs/complexity, and exploding information streams that often do not effectively reach the front lines hinder the ability to choose optimal treatment decisions over time.
TL;DR: It is shown that the no-cycle assumption can be removed if all subsystems of the follower have the same nominal dynamics and by directly making use of the property of the internal model, this technical note provides a more straightforward proof.
Abstract: The cooperative output regulation problem for linear uncertain multi-agent systems was studied in via an internal model approach under the assumption that the information graph of the system contains no cycle. In this technical note, we further show that the no-cycle assumption can be removed if all subsystems of the follower have the same nominal dynamics. Moreover, by directly making use of the property of the internal model, we provide a more straightforward proof than the one in in that we don't need to verify the satisfaction of certain matrix equations.
TL;DR: This note studies event-triggered control of Multi-Agent Systems with first-order integrator dynamics by considering limited communication capabilities through strict peer-to-peer non-continuous information exchange and provides both a decentralised control law and a decentralising communication policy.
Abstract: This note studies event-triggered control of Multi-Agent Systems (MAS) with first-order integrator dynamics. It extends previous work on event-triggered consensus by considering limited communication capabilities through strict peer-to-peer non-continuous information exchange. The approach provides both a decentralised control law and a decentralised communication policy. Communication events require no global information and are based only on local state errors; agents do not require a global sampling period or synchronous broadcasting as in sampled-data approaches. The proposed decentralised event-triggered control technique guarantees that the inter-event times for each agent are strictly positive. Finally, the ideas in this note are used to consider the practical scenario where agents are able to exchange only quantised measurements of their states.
TL;DR: This paper studies the smart control issue for an autonomous microgrid in order to maintain the secure voltages as well as maximize economic and environmental benefits.
Abstract: This paper studies the smart control issue for an autonomous microgrid in order to maintain the secure voltages as well as maximize economic and environmental benefits. A control scheme called as multi-agent based hierarchical hybrid control is proposed versus the hierarchical control requirements and hybrid dynamic behaviors of the microgrid. The control scheme is composed of an upper level energy management agent, several middle level coordinated control agents and many lower level unit control agents. The goals of smart control are achieved by designed control strategies. The simulations are given to demonstrate the effectiveness of proposed smart control for an autonomous microgrid.
TL;DR: In this article, a distributed leader-follower control algorithm for linear multi-agent systems based on output regulation theory and internal model principle is presented, where a leader is followed as an exosystem and the identical agents can track an active leader with different dynamics and unmeasurable variables.
Abstract: SUMMARY
In this paper, distributed leader–follower control algorithms are presented for linear multi-agent systems based on output regulation theory and internal model principle. By treating a leader to be followed as an exosystem, the proposed framework can be used to generalize existing multi-agent coordination solutions to allow the identical agents to track an active leader with different dynamics and unmeasurable variables. Moreover, the obtained results for multi-agent coordination control are an extension of previous work on centralized and decentralized output regulation to a distributed control context. Necessary and sufficient conditions for the distributed output regulation problem are given. Finally, distributed output regulation of some classes of multi-agent systems with switching interconnection topologies are discussed via both static and dynamic feedback. Copyright © 2011 John Wiley & Sons, Ltd.
17 Oct 2013
TL;DR: It is shown how to learn competent behaviors when a model of the game dynamics is available or when it is not, when full state information is given to the agent or just subjective observations, when learning is interactive or in batch-mode, and for a number of different learning algorithms, including reinforcement learning and evolutionary search.
Abstract: We propose a powerful new tool for conducting research on computational intelligence and games. `PyVGDL' is a simple, high-level description language for 2D video games, and the accompanying software library permits parsing and instantly playing those games. The streamlined design of the language is based on defining locations and dynamics for simple building blocks, and the interaction effects when such objects collide, all of which are provided in a rich ontology. It can be used to quickly design games, without needing to deal with control structures, and the concise language is also accessible to generative approaches. We show how the dynamics of many classical games can be generated from a few lines of PyVGDL. The main objective of these generated games is to serve as diverse benchmark problems for learning and planning algorithms; so we provide a collection of interfaces for different types of learning agents, with visual or abstract observations, from a global or first-person viewpoint. To demonstrate the library's usefulness in a broad range of learning scenarios, we show how to learn competent behaviors when a model of the game dynamics is available or when it is not, when full state information is given to the agent or just subjective observations, when learning is interactive or in batch-mode, and for a number of different learning algorithms, including reinforcement learning and evolutionary search.
TL;DR: Existing trust models from a game theoretic perspective are analyzed to highlight the special implications of including human beings in an MAS, and a possible research agenda to advance the state of the art in this field is proposed.
Abstract: In open and dynamic multiagent systems (MASs), agents often need to rely on resources or services provided by other agents to accomplish their goals. During this process, agents are exposed to the risk of being exploited by others. These risks, if not mitigated, can cause serious breakdowns in the operation of MASs and threaten their long-term wellbeing. To protect agents from the uncertainty in the behavior of their interaction partners, the age-old mechanism of trust between human beings is re-contexted into MASs. The basic idea is to let agents self-police the MAS by rating each other on the basis of their observed behavior and basing future interaction decisions on such information. Over the past decade, a large number of trust management models were proposed. However, there is a lack of research effort in several key areas, which are critical to the success of trust management in MASs where human beings and agents coexist. The purpose of this paper is to give an overview of existing research in trust management in MASs. We analyze existing trust models from a game theoretic perspective to highlight the special implications of including human beings in an MAS, and propose a possible research agenda to advance the state of the art in this field.
TL;DR: This manuscript provides the comprehensive overview of methodologies, architectures and applications of agents in industrial domain from early nineties up to present and gives an outlook of the current trends as well as challenges and possible future application domains of industrial agents.
Abstract: Industrial agents technology leverages the benefits of multiagent systems, distributed computing, artificial intelligence techniques and semantics in the field of production, services and infrastructure sectors, providing a new way to design and engineer control solutions based on the decentralization of control over distributed structures. The key drivers for this application are the benefits of agent-based industrial systems, namely in terms of robustness, scalability, reconfigurability and productivity, all of which translate to a greater competitive advantage. This manuscript monitors the chronology of research and development of the industrial applications of multiagent and holonic systems. It provides the comprehensive overview of methodologies, architectures and applications of agents in industrial domain from early nineties up to present. It also gives an outlook of the current trends as well as challenges and possible future application domains of industrial agents.
TL;DR: Control problems such as multirobot control, distributed intelligence, swarm intelligence, distributed decision, distributed cognition, congestion control in networks, collective motion in biology, oscillator synchronization in physics, parallelization in optimization theory, distributed estimation, cooperative estimation, equilibria in economics, social interaction modeling, and game theory may be analyzed under the theory of interconnected dynamic systems.
Abstract: Control problems such as multirobot control, distributed intelligence, swarm intelligence, distributed decision, distributed cognition, congestion control in networks, collective motion in biology, oscillator synchronization in physics, parallelization in optimization theory, distributed estimation, cooperative estimation, equilibria in economics, social interaction modeling, and game theory may be analyzed under the theory of interconnected dynamic systems. Those topics have several overlapping research communities; for that reason they are characterized by different definitions and a variety of approaches ranging from rigorous mathematical analysis to trial-and-error experimental study or emulation by observation of natural phenomena. The areas involved concern robotics, dynamic systems, computer science, signal theory, biology, economics, and mathematics. A shared taxonomy is missing; for example, dynamic systems can be identified in robots, agents, nodes, processors, and entities. An ensemble is called a group, network, platoon, swarm, team, and cluster, and the algorithms are defined as controllers, protocols, and dynamics. In the following, the term agent is used to denote the single dynamic system and network or collective the ensemble.
TL;DR: A detailed mean-square error analysis is performed and it is established that all agents are able to converge to the same Pareto optimal solution within a sufficiently smallmean-square-error (MSE) bound even for constant step-sizes.
Abstract: We consider solving multi-objective optimization problems in a distributed manner by a network of cooperating and learning agents. The problem is equivalent to optimizing a global cost that is the sum of individual components. The optimizers of the individual components do not necessarily coincide and the network therefore needs to seek Pareto optimal solutions. We develop a distributed solution that relies on a general class of adaptive diffusion strategies. We show how the diffusion process can be represented as the cascade composition of three operators: two combination operators and a gradient descent operator. Using the Banach fixed-point theorem, we establish the existence of a unique fixed point for the composite cascade. We then study how close each agent converges towards this fixed point, and also examine how close the Pareto solution is to the fixed point. We perform a detailed mean-square error analysis and establish that all agents are able to converge to the same Pareto optimal solution within a sufficiently small mean-square-error (MSE) bound even for constant step-sizes. We illustrate one application of the theory to collaborative decision making in finance by a network of agents.
TL;DR: This paper derives a necessary and sufficient condition for the leaderless consensus problem under a general directed graph and introduces a distributed filter for each agent or follower to show the effectiveness of the proposed control algorithms.
Posted Content•
TL;DR: In this article, a step-by-step introduction to the mathematical modelling based on a mesoscopic description and the construction of efficient simulation algorithms by Monte Carlo methods is provided, which can shed light on significant problems of the natural sciences as well as our daily lives.
Abstract: The description of emerging collective phenomena and self-organization in systems composed of large numbers of individuals has gained increasing interest from various research communities in biology, ecology, robotics and control theory, as well as sociology and economics. Applied mathematics is concerned with the construction, analysis and interpretation of mathematical models that can shed light on significant problems of the natural sciences as well as our daily lives. To this set of problems belongs the description of the collective behaviours of complex systems composed by a large enough number of individuals. Examples of such systems are interacting agents in a financial market, potential voters during political elections, or groups of animals with a tendency to flock or herd. Among other possible approaches, this book provides a step-by-step introduction to the mathematical modelling based on a mesoscopic description and the construction of efficient simulation algorithms by Monte Carlo methods. The arguments of the book cover various applications, from the analysis of wealth distributions, the formation of opinions and choices, the price dynamics in a financial market, to the description of cell mutations and the swarming of birds and fishes. By means of methods inspired by the kinetic theory of rarefied gases, a robust approach to mathematical modelling and numerical simulation of multi-agent systems is presented in detail. The content is a useful reference text for applied mathematicians, physicists, biologists and economists who want to learn about modelling and approximation of such challenging phenomena.
TL;DR: This article presents a multi-agent system model for virtual power plants, a new power plant concept in which generation no longer occurs in big installations, but is the result of the cooperation of smaller and more intelligent elements.
Abstract: Recent technological advances in the power generation and information technologies areas are helping to change the modern electricity supply system in order to comply with higher energy efficiency and sustainability standards. Smart grids are an emerging trend that introduce intelligence in the power grid to optimize resource usage. In order for this intelligence to be effective, it is necessary to retrieve enough information about the grid operation together with other context data such as environmental variables, and intelligently modify the behavior of the network elements accordingly. This article presents a multi-agent system model for virtual power plants, a new power plant concept in which generation no longer occurs in big installations, but is the result of the cooperation of smaller and more intelligent elements. The proposed model is not only focused on the management of the different elements, but includes a set of agents embedded with artificial neural networks for collaborative forecasting of disaggregated energy demand of domestic end users, the results of which are also shown in this article.