Consensus problems of first-order dynamic multi-agent systems with multiple time delays

Consensus problems of first-order multi-agent systems with multiple time delays are investigated in this paper. We discuss three cases: 1) continuous, 2) discrete, and 3) a continuous system with a proportional plus derivative controller. In each case, the system contains simultaneous communication and input time delays. Supposing a dynamic multi-agent system with directed topology that contains a globally reachable node, the sufficient convergence condition of the system is discussed with respect to each of the three cases based on the generalized Nyquist criterion and the frequency-domain analysis approach, yielding conclusions that are either less conservative than or agree with previously published results. We know that the convergence condition of the system depends mainly on each agent’s input time delay and the adjacent weights but is independent of the communication delay between agents, whether the system is continuous or discrete. Finally, simulation examples are given to verify the theoretical analysis.     

知识共享与间接互惠:基于计算实验方法的研究

团队中,个体知识共享行为的回报往往来自第三方而非受助者,而以直接互惠为视角的知识共享研究却不足以解释此类现象。采用基于计算机仿真实验方法,通过赋予Agent记忆、推理、决策和沟通等能力,研究受表型背叛、组织信任氛围及声誉传播机制影响下,间接互惠机制对团队大范围知识共享行为的维系机理。研究发现,由于辨别者的存在,即使无条件共享者会成为绝对多数,隐藏者并不会大量侵入群体。这意味着间接互惠机制维系了团队大范围知识共享行为。在低的组织信任氛围水平下,辨别者和隐藏者虽然能够共存,团队内的知识共享行为却没有出现。即使只有少数成员能观察到周围同事的知识共享行为,只要声誉信息能够有效传播,间接互惠对广泛知识共享行为的维系作用就能够发挥。     

Scalable and Transferable Reinforcement Learning for Multi-Agent Mixed Cooperative–Competitive Environments Based on Hierarchical Graph Attention

Most previous studies on multi-agent systems aim to coordinate agents to achieve a common goal, but the lack of scalability and transferability prevents them from being applied to large-scale multi-agent tasks. To deal with these limitations, we propose a deep reinforcement learning (DRL) based multi-agent coordination control method for mixed cooperative–competitive environments. To improve scalability and transferability when applying in large-scale multi-agent systems, we construct inter-agent communication and use hierarchical graph attention networks (HGAT) to process the local observations of agents and received messages from neighbors. We also adopt the gated recurrent units (GRU) to address the partial observability issue by recording historical information. The simulation results based on a cooperative task and a competitive task not only show the superiority of our method, but also indicate the scalability and transferability of our method in various scale tasks.     

基于多Agent的宏观经济智能预测决策支持系统研究

在分析智能决策支持系统的发展基础上，将人工智能思想，特别是Agent应用于构建智能宏观经济预测决策支持系统。提出面向任务的Agent设计思想，以任务为核心设计多种类型Agent共同完成预测决策任务，给出基于多Agent的系统组成结构，并给出系统中不同类型Agent的具体实现结构，同时探讨了整个系统的运行机制以及系统中多Agent之间的合作。整个系统的构建过程也是面向Agent的程序设计的过程。     

不确定环境下多自主体系统的分布式估计与控制

多自主体系统是当前系统控制界研究的热点问题. 在实际中, 自主体系统通常并不是在理想的环境下执行任务, 而是面临多源头、多层次和多变化的各类不确定性因素的影响. 它们通过在微观层面上影响各自主体决策的正确性, 从而在宏观上对自主体系统的整体行为产生显著影响. 不确定性因素和多自主体系统分布式信息架构交互耦合, 给系统的设计与分析带来本质性困难. 本文围绕分布式估计与分布式控制问题, 研究在随机通信噪声、数据丢失、量化和系统未知结构参数等不确定因素影响下, 如何为各自主体设计更加鲁棒、更加有效的分布式估计算法及分布式控制律, 以实现全局估计与控制目标, 并对闭环系统性能进行系统分析.     

TaskSelSim: a model of the self-organization of the division of labour in honeybees

We here describe a novel multi-agent model of the decentralized task selection process in honeybee colonies. The model is based on individual behavioural programs represented as hierarchical finite state automatons. We successfully simulated empirical experiments (with real honeybees) using the model. We shed more light on task specialization and polyethism in honeybees by modelling the heterogeneous environment that emerges over time inside the colony and by considering the changing differences in stimuli strength in each small area of the hive. Our simulation experiments presented here investigate the model's reaction to changing values in crucial model parameters and investigate the stability of the predicted colony behaviours.     

Multi-agent system model with mixed coupling topologies for pattern formation and formation splitting

In this article, a new type of multi-agent system model with mixed coupling topologies is proposed for realizing pattern formations with specific geometric shapes and formation splitting. The interactions among individual agents are assumed to be universally repulsive and selectively attractive. By designing the form of attractive coupling matrix, one can obtain a variety of formations with specific shapes in the system through self-assembly of agents. Both symmetric coupling case and asymmetric coupling case are considered. Analysis and simulation results show symmetric ones result in convergent dynamics to steady-state formations, whereas, for asymmetric case, the system exhibits complex dynamic behaviours, including collective rotation and chaotic motion. By breaking the graph defined by attractive couplings into disjoint subgraphs, one can make the formation of agents to split into small sizes. The results are relevant for the design of coordination and cooperative control for multi-agent systems.     

Tracking of a third-order maneuvering target under an arbitrary topology

We explore the tracking problem of a maneuvering target. Tracking agents with third-order kinematics can communicate with each other via wireless network. The communication network topology is arbitrary rather than switches among several fixed topologies. The information sharing and interaction among agents are position, velocity, and acceleration. Some sufficient conditions of tracking strategy have been proposed. Finally, a numerical example is employed to demonstrate the effectiveness of proposed tracking strategy.     

Consensus protocol for heterogeneous multi-agent systems: A Markov chain approach

This paper deals with the consensus problem for heterogeneous multi-agent systems. Different from most existing consensus protocols, we consider the consensus seeking of two types of agents, namely, active agents and passive agents. The objective is to directly control the active agents such that the states of all the agents would achieve consensus. In order to obtain a computational approach, we subtly introduce an appropriate Markov chain to cast the heterogeneous systems into a unified framework. Such a framework is helpful for tackling the constraints from passive agents. Furthermore, a sufficient and necessary condition is established to guarantee the consensus in heterogeneous multi-agent systems. Finally, simulation results are provided to verify the theoretical analysis and the effectiveness of the proposed protocol.     

Average consensus of multi-agent systems with communication time delays and noisy links

In this paper,we consider the average-consensus problem with communication time delays and noisy links.We analyze two different cases of coupling topologies:fixed and switching topologies.By utilizing the stability theory of the stochastic differential equations,we analytically show that the average consensus could be achieved almost surely with the perturbation of noise and the communication time delays even if the time delay is time-varying.The theoretical results show that multi-agent systems can tolerate relatively large time delays if the noise is weak,and they can tolerate relatively strong noise if the time delays are low.The simulation results show that systems with strong noise intensities yield slow convergence.