基于改进头姿估计方法的机器人轮椅交互控制

针对现有迭代最近点（ICP）头姿估计算法存在迭代次数偏多且易陷于局部最优、而随机森林（RF）头姿估计算法准确性和稳定性不高的问题,提出一种新的头姿估计改进方法,并基于该改进方法构建机器人轮椅实时交互控制接口．首先,分析现有迭代最近点头姿算法与随机森林头姿算法在准确性、实时性及稳定性方面存在的问题,并提出一种新的基于随机森林与迭代最近点算法融合的头姿估计改进方法;其次,为实现头姿估计到机器人轮椅交互控制的无缝连接,建立基于传统机器人轮椅操纵杆的头部姿态运动空间映射;最后,在基于标准头姿数据库分析改进头姿估计方法性能的基础上,构建机器人轮椅实验平台并规划运动轨迹,以进一步验证基于改进头姿估计方法的人机交互接口在机器人轮椅实时控制方面的有效性．实验结果表明,改进后的头姿估计方法较传统迭代最近点算法减少了迭代次数且避免了陷于局部最优,在仅增加少量运算时间的基础上,其准确性和稳定性都优于传统随机森林算法;同时,基于改进头姿估计方法的人机交互接口亦能实时平稳地控制机器人轮椅沿既定的轨迹运动．     

工业机器人空间曲线实时轨迹规划算法

本文设计了一种六轴机械臂沿空间曲线行走的控制算法。算法使机器人对输入空间曲线计算出各关节的电机控制命令使机器人沿曲线运动。算法中推导出机器人逆运动学各关节解析解,使机械臂控制器计算过程由解方程组简化为代数运算。对类似结构的机械臂都可以使用,具有一定的通用性。使用Matlab对算法进行编程实现,从而验证了文章所设计算法的正确性。     

具有轨迹约束的操作臂避障

该文提出一种基于多Agent的操作臂控制算法,借助于拟合度的概念,使操作臂末端点沿任一给定轨迹运行,进而可以实现避障功能。     

改进末端跟随运动的超冗余蛇形臂机器人运动学逆解

针对超冗余蛇形臂机器人运动学逆解中计算量大、超关节极限和位形偏移量大的问题,提出了一种改进末端跟随运动的逆解算法．在末端跟随法中引入蛇形臂弯曲角度的约束,调整关节位置的更新方式,使关节在蛇形臂轴线上运动．通过依次更新关节的空间位置,将超冗余多节蛇形臂的运动学逆解转化为2自由度单节蛇形臂的运动学逆解．仿真分析了蛇形臂机器人在基座移动和基座固定条件下的轨迹跟踪效果,对比了同一目标位置下不同方法的性能．结果表明,改进后的算法能保证蛇形臂的弯曲角度不超过给定范围,关节的运动量从末端到基座依次减小,机器人的运动更协调;与基于雅可比矩阵的数值法和现有启发式方法相比,该方法运算量降低,机器人整体位形偏移量减小,能用于蛇形臂机器人的实时控制．     

一种基于DTW-GMM的机器人多机械臂多任务协同策略

为了控制机器人完成复杂的多臂协作任务, 提出了一种基于动态时间规整?高斯混合模型(Dynamic time warping-Gaussian mixture model, DTW-GMM)的机器人多机械臂多任务协同策略. 首先, 针对机器人示教时轨迹时间长短往往存在较大差异的问题, 采用动态时间规整方法来统一时间的变化; 其次, 基于动态时间规整的多机械臂示教轨迹, 采用高斯混合模型对轨迹的特征进行提取, 并以某一机械臂的位置空间矢量作为查询向量, 基于高斯混合回归泛化输出其余机械臂的执行轨迹; 最后, 在Pepper仿人机器人平台上验证了所提出的多机械臂协同策略, 基于DTW-GMM算法控制机器人完成了双臂协作搬运任务和汉字轨迹的书写任务. 提出的基于DTW-GMM算法的多任务协同策略简单有效, 可以利用反馈信息实时协调各机械臂的任务, 在线生成平滑的协同轨迹, 控制机器人完成复杂的协作操作.     

基于蚁群算法及博弈论的多Agent路径规划算法

针对多Agent路径规划问题,提出了一个两阶段的路径规划算法。首先,利用改进的蚁群算法来为每个Agent规划出一条从起始点到目标点,不与环境中静态障碍物碰撞的最优路径。在蚁群算法的改进中引入反向学习方法来对蚂蚁位置进行初始化分布,提高了算法的全局搜索能力;利用粒子群算法中的自适应惯性权重因子来调节信息素强度Q值,使其自适应地变化,避免陷入局部最优;对信息素挥发因子ρ进行调节,提高算法的迭代速度。其次,若多Agent之间存在动态碰撞,利用博弈论构建多Agent之间的动态避障模型,并利用虚拟行动法来解决博弈的求解问题及多Nash均衡的选择问题,确保每个Agent能够快速学习到最优Nash均衡。仿真实验结果表明改进蚁群算法与传统蚁群算法相比在搜索精度与搜索速度上有明显的提高,与Mylvaganam的多Agent动态避障算法相比,所提算法减小了路径总长度并提高了收敛速度。     

混凝土泵车控制系统优化算法的设计

以国内某公司的四臂混凝土泵车作为控制对象,运用机器人运动学,建立其运动学模型,基于该模型给出泵车运行轨迹优化问题。考虑系统利用PLC控制器实现,因此提出了一种新的快速简便的泵车的控制系统优化算法。经Matlab仿真实验,该算法运算时间少,明显比工人手工操作能更快更好的到达目的点。     

基于Agent的MCU负载均衡方法

针对视频会议系统中多点控制单元(MCU)负载均衡问题,提出一种基于Agent的MCU负载均衡方法和一种具有较低MCU级联的组优先负载分配算法。该方法通过负载Agent和负载均衡Agent收集区域内MCU的动态负载信息并执行负载均衡操作,使MCU的负载得到平衡。实验结果证明该方法与FRFA算法相比,传输时延约减少8%。     

基于BDI框架的多Agent动态协作模型与应用研究

近年来，多Agent学习已经成为人工智能和机器学习研究方向发展最迅速的领域之一．将强化学习和BDI思维状态模型相结合，形成针对多Agent的动态协作模型．在此模型中，个体最优化概念失去其意义，因为每个Agent的回报，不仅取决于自身，而且取决于其它Agent的选择．模型采用AFS神经网络对输入状态空间进行压缩，提高强化学习的收敛速度．与此同时，利用模拟退火算法启发性地指明动作空间搜索方向，使其跳出局部最小点，避免迭代步数的无限增长．理论分析和在机器人足球领域的成功应用，都证明了基于BDI框架的多Agent动态协作模型的有效性。     

输入约束不确定系统的点对点迭代学习控制与优化

为解决工业过程中机械臂等特殊重复运行系统的输出在有限时间内无需实现全轨迹跟踪,仅需跟踪期望轨迹上某些特殊关键点的控制问题,针对线性时不变离散系统提出一种基于范数最优的点对点迭代学习控制算法.通过输入输出时间序列矩阵模型变换构建综合性多目标点性能指标函数,求解二次型最优解得到优化迭代学习控制律,同时给出模型标称和不确定情形下最大奇异值形式鲁棒控制算法收敛的充分条件,并进一步推广得到输入约束系统优化控制算法的收敛性结果,最后在三轴龙门机器人模型上验证算法的有效性.     

基于串联结构的分布式模型预测控制

分布式模型预测控制(Distributed model predictive control, DMPC)是一类用于多输入多输出的大规模系统的控制方式.每个智能体通过相互协作完成整个系统的控制. 已有的分布式预测控制算法可以划分为迭代式算法和非迭代算法:迭代算法在迭代到收敛情况下,具有集中式预测控制(Centralized model predictive control, CMPC)算法的性能,但迭 代次数过多,子系统间通信量大;非迭代算法不需要迭代,但性能有一定损失.本文提出了一种基于串联结构的非迭代分布式预测控 制算法.本文算法在串联结构系统中可以有效减少计算量,并结合氧化铝碳分解(Alumina continuous carbonation decomposition process, ACCDP)这一串联过程,通过仿真验证了算 法的有效性;同时分析了算法运用在串联结构下的性能并证明了其稳定性.     

A Heuristic Approach to the Inverse Differential Kinematics Problem

Inversion of the kinematics of manipulators is one of the central problems in the field of robot arm control. The iterative use of inverse differential kinematics is a popular method of solving this task. Normally the solution of the problem requires a complex mathematical apparatus. It involves methods for solving equation systems as well as algorithms for optimization. In this paper we introduce a naïve heuristic method which works without the need for complex mathematical algorithms. This method forms a simple basis for the more sophisticated control procedures of our robot manipulator (JANUS).     

Formation control for multi‐agent systems through an iterative learning design approach

This paper deals with formation control problems for multi‐agent systems by using iterative learning control (ILC) design approaches. Distributed formation ILC algorithms are presented to enable all agents in directed graphs to achieve the desired relative formations perfectly over a finite‐time interval. It is shown that not only asymptotic stability but also monotonic convergence of multi‐agent formation ILC can be accomplished, and the convergence conditions in terms of linear matrix inequalities can be simultaneously established. The derived results are also applicable to multi‐agent systems that are subject to stochastic disturbances and model uncertainties. Furthermore, the feasibility of convergence conditions and the effect of communication delays are discussed for the proposed multi‐agent formation ILC algorithms. Simulation results are given for uncertain multi‐agent systems to verify the theoretical study. Copyright © 2012 John Wiley & Sons, Ltd.     

基于免疫算法的多匝道MF-AILC控制方法

针对高速公路交通系统的重复性和周期性,提出基于免疫算法的高速公路多匝道无模型自适应迭代学习控制（MF—AILC）联合协调控制,以高速公路最大流量、全局最小行程时间和入口平均等待时间三者为目标函数,用免疫算法对多匝道协调迭代学习控制器进行优化。仿真结果表明,与无控制情况下进行比较,该控制方法具有良好的控制效果和较强的鲁棒性。     

Bounded Consensus Algorithms for Multi‐Agent Systems in Directed Networks

In this paper, the consensus problem is investigated via bounded controls for the multi‐agent systems with or without communication. Based on the nested saturation method, the saturated control laws are designed to solve the consensus problem. Under the designed saturated control laws, the transient performance of the closed‐loop system can be improved by tuning the saturation level. First of all, asymptotical consensus algorithms with bounded control inputs are proposed for the multi‐agent systems with or without communication delays. Under these consensus algorithms, the states’ consensus can be achieved asymptotically. Then, based on a kind of novel nonlinear saturation functions, bounded finite‐time consensus algorithms are further developed. It is shown that the states’ consensus can be achieved in finite time. Finally, two examples are given to verify the efficiency of the proposed methods.     

High‐precision formation control of nonlinear multi‐agent systems with switching topologies: A learning approach

Arbitrary high precision is considered one of the most desirable control objectives in the relative formation for many networked industrial applications, such as flying spacecrafts and mobile robots. The main purpose of this paper is to present design guidelines of applying the iterative schemes to develop distributed formation algorithms in order to achieve this control objective. If certain conditions are met, then the control input signals can be learned by the developed algorithms to accomplish the desired formations with arbitrary high precision. The systems under consideration are a class of multi‐agent systems under directed networks with switching topologies. The agents have discrete‐time affine nonlinear dynamics, but their state functions do not need to be identical. It is shown that the learning processes resulting from the relative output formation of multi‐agent systems can converge exponentially fast with the increase of the iteration number. In particular, this work induces a distributed algorithm that can simultaneously achieve the desired relative output formation between agents and regulate the movement of multi‐agent formations as desired along the time axis. The illustrative numerical simulations are finally performed to demonstrate the effectiveness and performance of the proposed distributed formation algorithms. Copyright © 2014 John Wiley & Sons, Ltd.     

Distributed iterative command governor schemes for interconnected linear systems

A novel distributed command governor (CG) supervision strategy relying on iterative optimization procedure is presented for multi‐agent interconnected linear systems subject to pointwise‐in‐time set‐membership coordination constraints. Unlike non‐iterative distributed CG schemes, here all agents undertake several optimization iterations and data exchange before arriving to the optimal solution. As a result, these methods are able to achieve Pareto‐optimal solutions not only in steady‐state conditions as the ones based on non‐iterative optimization procedures but also during transients and are not hampered by the presence of undesirable Nash‐equilibria or deadlock situations. The main properties of the method are fully investigated and in particular its optimality, stability, and feasibility properties rigorously proved. A final example is presented where the proposed distributed solution is contrasted with existing centralized and distributed non‐iterative CG solutions. Copyright © 2017 John Wiley & Sons, Ltd.     

超立方体多处理机系统中基于扩展最优通路矩阵的容错路由

该文在高峰等文章的基础上，提出了针对超立方体结构多处理机系统的扩展最优通路矩阵（Extended Optimal Path Matrices,EOPMs）的概念，并给出了一个建立EIPMs的算法和基于EOPMs的容错路由算法，证明了基于EOPMs的容错路由算法是基于扩展安全向量（ESVs）^[13]和基于最优通路矩阵（OPMs)^[14]容错路由算法的扩展，与原文相比，该算法的存储开销与OPMs,相同，但记录的最优通路的信息，包含了原文所记录的最优通路的信息，使搜索最优通路的能力比它们有进一步的提高。     

基于深度学习的多机械手轨迹规划系统设计

目前提出的多机械手轨迹规划系统路径规划精准度低,避障能力差。基于深度学习对多机械手的规划系统进行设计,通过研究传统系统中存在精确度、智能性不足的缺点,在设计的系统分别引入了相应的解决条件,在硬件结构的设计中本文应用ISL-320型号的伺服电机提升多机械手的动力功能,应用SKT64系列的芯片提升多机械手的路径精准度;在应用程序设计上应用拟合算法与叠加算法对规划路径中的节点精准运算,在提升系统整体精准度的同时提升了系统的智能程度。实验结果表明,基于深度学习的多机械手轨迹规划系统路径与标准路径十分接近,说明该方法的规划精准度较高,避障能力得到有效增强。     

Wavelet fuzzy LVQ based speaker verification system

This work evaluates the performance of speaker verification system based on Wavelet based Fuzzy Learning Vector Quantization (WLVQ) algorithm. The parameters of Gaussian mixture model (GMM) are designed using this proposed algorithm. Mel Frequency Cepstral Coefficients (MFCC) are extracted from the speech data and vector quantized through Wavelet based FLVQ algorithm. This algorithm develops a multi resolution codebook by updating both winning and nonwinning prototypes through an unsupervised learning process. This codebook is used as mean vector of GMM. The other two parameters, weight and covariance are determined from the clusters formed by the WLVQ algorithm. The multi resolution property of wavelet transform and ability of FLVQ in regulating the competition between prototypes during learning are combined in this algorithm to develop an efficient codebook for GMM. Because of iterative nature of Expectation Maximization (EM) algorithm, the applicability of alternative training algorithms is worth investigation. In this work, the performance of speaker verification system using GMM trained by LVQ, FLVQ and WLVQ algorithms are evaluated and compared with EM algorithm. FLVQ and WLVQ based training algorithms for modeling speakers using GMM yields better performance than EM based GMM.