动态环境中的多机器人协同搬运

在多机器人协同搬运过程中,针对传统的强化学习算法仅使用数值分析却忽略了推理环节的问题,将多机器人的独立强化学习与“信念-愿望-意向”（BDI）模型相结合,使得多机器人系统拥有了逻辑推理能力,并且,采用距离最近原则将离障碍物最近的机器人作为主机器人,并指挥从机器人运动,提出随多机器人系统位置及最近障碍物位置变化的评价函数,同时将其与基于强化学习的行为权重结合运用,在多机器人通过与环境不断交互中,使行为权重逐渐趋向最佳。仿真实验表明,该方法可行,能够成功实现协同搬运过程。     

一种局部动态环境下的避障算法

提出一种基于传统VFH避障算法的增强形式,称为VFH＃算法。这种算法对VFH算法第一层进行局部环境预测,重新确定静态动态栅格,为后几层的最优选择提供准确的参数,利用这种方法,机器人能在局部动态环境下选择较化的行进方向。     

基于平行控制的异构多Agent机器人系统控制方法

复杂系统控制过程往往需要人综合控制系统、仿真系统,以及人在回路综合实现。构建基于多agent机器人系统及其实现策略,是多机器人系统研究和解决现实问题的热点。本文基于多部件综合作业机器人项目,针对不同作业机器人,提出一种异构多Agent机器人系统控制方法,此方法引入ACP理论的思想,将平行控制方法应用于机器人综合作业平台,通过人工系统与实际系统的虚实互动,使电脑的计算能力与人脑的灵活能力有效结合,实现了一个复杂系统的合理管控,为多机器人协作完成实际复杂控制系统提供了一种保证系统可靠性基础上提高效率的可供借鉴的理论和实现方法。     

不确定环境下多机器人的动态编队控制

提出了一种不确定环境下多机器人的动态编队控制方法．通过队形参数矩阵确立多机器人之间的相对 位置关系,将全局队形控制问题转化为跟随机器人离轴点对虚机器人（与领航机器人运动方向一致,且对领航机器 人保持期望的相对距离和观测方位角）离轴点的跟踪．基于建立的跟随机器人和领航机器人之间的误差跟踪系统模 型设计相应控制律实现队形保持,并提出了防止机器人与障碍物及其它机器人碰撞的避障策略．仿真结果表明了所 提方法的可行性和有效性．     

多机器人动态编队的强化学习算法研究

在人工智能领域中，强化学习理论由于其自学习性和自适应性的优点而得到了广泛关注．随着分布式人工智能中多智能体理论的不断发展，分布式强化学习算法逐渐成为研究的重点．首先介绍了强化学习的研究状况，然后以多机器人动态编队为研究模型，阐述应用分布式强化学习实现多机器人行为控制的方法．应用SOM神经网络对状态空间进行自主划分，以加快学习速度；应用BP神经网络实现强化学习，以增强系统的泛化能力；并且采用内、外两个强化信号兼顾机器人的个体利益及整体利益．为了明确控制任务，系统使用黑板通信方式进行分层控制．最后由仿真实验证明该方法的有效性．     

基于群集行为的分布式多无人机编队动态避障控制

针对无人机编队保持和动态障碍物规避控制问题,本文提出了一种新的基于群集行为的分布式多无人机编队控制和避障控制算法.首先考虑了由机间气流等因素带来的干扰,基于吸引/排斥势场和一致性方法,设计了分布式无人机编队的队形保持控制算法,对编队内无人机之间的距离进行控制.进一步考虑外部移动障碍对无人机编队的影响,引入了排斥势场产生...     

基于强化学习的六足机器人动态避障研究

针对局部可观测的非线性动态地震环境下,六足机器人采用传统算法进行动态避障时易出现算法不稳定的情况.运用了基于双重深度Q网络(DDQN)的决策方式,通过传感器数据输入卷积神经网络(CNN)并结合强化学习的策略,下达命令到六足机器人,控制输出决策动作,实现机器人动态避障.将系统的环境反馈与决策控制直接形成闭环,通过最大化机...     

多机器人任意队形分布式控制研究

本文针对多智能体协作完成特定任务时难以在全自主控制的前提下协作形成任意队 形和队形向量不易确定的问题，通过由各智能体自主简单的确定自己的队形向量，从理论上 扩展基于队形向量的队形控制原理以生成任意队形，改进机器人的运动方式以提高收敛速度 ，提出一种快速收敛的机器人部队任意队形分布式控制算法．为了解决智能体机器人之间的 冲突问题，提出了一个通信协调模型．仿真实验和实际机器人实验均表明了算法的可行性和 有效性．     

面向多机器人环境中动态异构任务的细粒度动作分配与调度方法

在多机器人环境中,具有不同能力的机器人相互协作以完成任务需求。现实情况下,这些任务动态发布,且具有不同的目标和紧急程度,因此需要为每个任务分解出的细粒度动作分配和调度合适的机器人来负责执行这些动作。现有的方法大多适用于静态和同构的任务分配场景,而针对动态异构任务的分配则大多采用独占式的分配策略,导致机器人频繁进入等待状态(即机器人处于被分配了任务到真正开始执行任务之间的闲置阶段)。由于任务存在不同的紧急程度和发布时间,这种分配方式将降低对更紧急任务的响应效率,同时导致更多的等待时间和更长的任务完成时间。针对该问题,提出了一种面向多机器人环境中动态异构任务的细粒度动作分配与调度方法。其中,分配与调度的对象是任务所分解出的细粒度的动作,且一个动作能够由机器人的一种能力承担。面对任务分解出的一组细粒度动作集合,本方法借鉴拍卖算法过程,根据机器人能力、状态及任务信息计算出机器人承担特定动作的最优分配方案。另外,在每一次新任务发布或某一机器人执行完动作时执行分配和调度过程,可以将处于普通任务等待状态的机器人调度至紧急任务,以保证紧急任务优先完成,且缩短机器人的总体等待时间。基于本方法,扩展实现了...     

时滞分布参数系统的移动传感器/执行器防碰撞控制

针对一类状态时滞的分布参数系统,考虑了传感器/执行器间的防碰撞问题和最大通讯距离的最小通讯能耗问题,以及系统的稳定性问题.利用抽象发展方程理论和Lyapunov稳定性方法,设计了一种基于时滞分布参数系统的智能体移动控制策略,包括输出反馈控制器和移动控制力.通过理论推导和仿真实验验证,文中设计的控制策略能够使得时滞分布参数系统是渐近稳定的,同时智能体在移动过程中是防碰撞的,也验证了智能体在最大通讯距离的最小能耗.     

Collision avoidance command governor for multi‐vehicle unmanned systems

In this paper, we consider the problem of coordinating a collection of autonomous unmanned vehicles while guaranteeing collision avoidance. Each vehicle is regulated by a local controller that ensures stability and provides desired path tracking performance in the absence of constraints. The fulfillment of coordination tasks (e.g., collision avoidance) and local constraints (e.g., input saturation constraints) is achieved through a command governor (CG) strategy that, whenever necessary, modifies the nominal paths of the vehicles. First, a centralized CG approach is proposed and fully analyzed. Then, a more interesting distributed implementation requiring low communication rates is discussed. Both approaches make use of a receding horizon strategy and require the on‐line solution of mixed‐integer optimization programs. Finally, an example is given for illustration purposes. Copyright © 2013 John Wiley & Sons, Ltd.     

Design of distributed PID-type dynamic matrix controller for fractional-order systems

With the continuous requirements for product quality and safety operation in industrial production, it is difficult to describe the complex large-scale processes with integer-order differential equations. However, the fractional differential equations may precisely represent the intrinsic characteristics of such systems. In this paper, a distributed PID-type dynamic matrix control method based on fractional-order systems is proposed. First, the high-order approximate model of integer order is obtained by utilising the Oustaloup method. Then, the step response model vectors of the plant is obtained on the basis of the high-order model, and the online optimisation for multivariable processes is transformed into the optimisation of each small-scale subsystem that is regarded as a sub-plant controlled in the distributed framework. Furthermore, the PID operator is introduced into the performance index of each subsystem and the fractional-order PID-type dynamic matrix controller is designed based on Nash optimisation strategy. The information exchange among the subsystems is realised through the distributed control structure so as to complete the optimisation task of the whole large-scale system. Finally, the control performance of the designed controller in this paper is verified by an example.     

动态事件触发机制下二阶多智能体系统完全分布式控制

本文研究了无向通信拓扑下二阶多智能体系统的一致性问题, 分别针对有领导者和无领导者的情形, 设计 了一类基于辅助动态变量的完全分布式事件触发控制策略, 该策略具有参数较少且易调等特点. 智能体自身的触 发函数满足条件时才向邻居广播自身的状态信息, 有效避免了连续通信, 减少了系统能量耗散. 每个智能体的控制 协议和触发函数都只用到自身的状态和邻居触发时刻的状态, 不涉及邻居的实时状态信息, 也不依赖通信拓扑网络 的任何全局信息. 利用代数图论以及Lyapunov稳定性理论, 证明在所提出的控制策略下, 二阶多智能体系统能够实 现渐近一致性, 且不存在Zeno行为. 仿真示例进一步验证了理论结果的有效性.     

Adaptive flexible performance function approach for dynamic obstacle avoidance to distributed formation tracking of range-constrained switched nonlinear multiagent systems

Prescribed performance control allows preselected transient and steady-state bounds of formation control performance. However, owing to the predetermined decaying property of performance functions, it is unavailable for the formation tracking problem with dynamic obstacle avoidance. Thus, it is essential to design a flexible performance function to accommodate the increasing formation error during dynamic obstacle avoidance. This article pays attention to the development of an adaptive flexible performance (AFP) function usable for the avoidance of dynamic obstacles with unknown velocities in the distributed formation framework. We develop an AFP-based distributed formation tracker design for range-constrained switched multi-input multi-output nonlinear multiagent systems with asynchronous switching. The AFP functions for ensuring graph connectivity and dynamic obstacle avoidance are derived by designing time-varying variables adjusted by an adaptive relaxation signal. The neural-network-based adaptive distributed formation tracker using the derived performance functions is constructed with the adaptive relaxation signal to ensure the stability of the closed-loop formation system, regardless of asynchronous switching and dynamic obstacles. The effectiveness of the proposed method is shown by simulation results.     

动态环境下改进蚁群算法的多Agent路径规划

针对动态环境下的多Agent路径规划问题,提出了一种改进的蚁群算法与烟花算法相结合的动态路径规划方法。通过自适应信息素强度值及信息素缩减因子来加快算法的迭代速度,并利用烟花算法来解决路径规划过程中的死锁问题,避免陷入局部最优。在多Agent动态避碰过程中,根据动态障碍物与多Agent之间的运行轨迹是否相交制定相应的避碰策略,并利用路径转变函数解决多Agent的正面碰撞问题。仿真实验表明,该方法优于经典蚁群算法,能够有效解决多Agent路径规划中的碰撞问题,从而快速找到最优无碰路径。     

Hierarchical multi-robot navigation and formation in unknown environments via deep reinforcement learning and distributed optimization

Compared with a single robot, Multi-robot Systems (MRSs) can undertake more challenging tasks in complex scenarios benefiting from the increased transportation capacity and fault tolerance. This paper presents a hierarchical framework for multi-robot navigation and formation in unknown environments with static and dynamic obstacles, where the robots compute and maintain the optimized formation while making progress to the target together. In the proposed framework, each single robot is capable of navigating to the global target in unknown environments based on its local perception, and only limited communication among robots is required to obtain the optimal formation. Accordingly, three modules are included in this framework. Firstly, we design a learning network based on Deep Deterministic Policy Gradient (DDPG) to address the global navigation task for single robot, which derives end-to-end policies that map the robot’s local perception into its velocity commands. To handle complex obstacle distributions (e.g. narrow/zigzag passage and local minimum) and stabilize the training process, strategies of Curriculum Learning (CL) and Reward Shaping (RS) are combined. Secondly, for an expected formation, its real-time configuration is optimized by a distributed optimization. This configuration considers surrounding obstacles and current formation status, and provides each robot with its formation target. Finally, a velocity adjustment method considering the robot kinematics is designed which adjusts the navigation velocity of each robot according to its formation target, making all the robots navigate to their targets while maintaining the expected formation. This framework allows for formation online reconfiguration and is scalable with the number of robots. Extensive simulations and 3-D evaluations verify that our method can navigate the MRS in unknown environments while maintaining the optimal formation.     

Recent Progress in Reinforcement Learning and Adaptive Dynamic Programming for Advanced Control Applications

Reinforcement learning (RL) has roots in dynamic programming and it is called adaptive/approximate dynamic programming (ADP) within the control community. This paper reviews recent developments in ADP along with RL and its applications to various advanced control fields. First, the background of the development of ADP is described, emphasizing the significance of regulation and tracking control problems. Some effective offline and online algorithms for ADP/adaptive critic control are displayed, where the main results towards discrete-time systems and continuous-time systems are surveyed, respectively. Then, the research progress on adaptive critic control based on the event-triggered framework and under uncertain environment is discussed, respectively, where event-based design, robust stabilization, and game design are reviewed. Moreover, the extensions of ADP for addressing control problems under complex environment attract enormous attention. The ADP architecture is revisited under the perspective of data-driven and RL frameworks, showing how they promote ADP formulation significantly. Finally, several typical control applications with respect to RL and ADP are summarized, particularly in the fields of wastewater treatment processes and power systems, followed by some general prospects for future research. Overall, the comprehensive survey on ADP and RL for advanced control applications has demonstrated its remarkable potential within the artificial intelligence era. In addition, it also plays a vital role in promoting environmental protection and industrial intelligence.