团队CGF行进中的队形控制方法

分布式虚拟环境中,团队CGF的行进问题是CGF研究的基本问题,而行进中的队形保持问题义是行进问题研究的重点,在提出的团队组织模型CTOM的基础上,给出了一种团队CGF行进中队形保持方法。     

社会网络中弱关系团队形成问题研究

随着在线社会网络的迅速发展,社会网络的团队形成问题逐渐成为研究热点。现有的社会网络中团队形成问题目标是寻找一个成员间沟通代价最小的团队。然而,实际应用中团队成员间的不紧密关系使得团队的观点多样化、多角度、无偏见,可以广泛应用于形成专家评审团队、大众评审团等。基于此需求,将社会学的弱关系概念引入团队形成问题中,提出了一种社会网络中弱关系团队形成问题。该问题旨在寻找成员间为弱关系,同时满足技能、经验值要求的一个团队,为NP-hard问题。提出了3类算法解决该问题,分别为贪心算法、精确算法、α-近似算法,每类算法有各自的特点与适用范围。利用ACM和DBLP两类真实的数据集进行实验,综合评估了各类算法的效率与求解质量,证明了提出算法的有效性。     

Handling Communication Restrictions and Team Formation in Congestion Games

There are many domains in which a multi-agent system needs to maximize a “system utility” function which rates the performance of the entire system, while subject to communication restrictions among the agents. Such communication restrictions make it difficult for agents that take actions to optimize their own “private” utilities to also help optimize the system utility. In this article we show how previously introduced utilities that promote coordination among agents can be modified to be effective in domains with communication restrictions. The modified utilities provide performance improvements of up to 75 over previously used utilities in congestion games (i.e., games where the system utility depends solely on the number of agents choosing a particular action). In addition, we show that in the presence of severe communication restrictions, team formation for the purpose of information sharing among agents leads to an additional 25 improvement in system utility. Finally, we show that agents’ private utilities and team sizes can be manipulated to form the best compromise between how “aligned” an agent’s utility is with the system utility and how easily an agent can learn that utility.     

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．     

基于强化学习的多机器人编队方法研究

介绍了国内外关于强化学习的研究现状,对应用Q-学习和神经网络来实现多机器人的自适应编队方法给出了详细的系统描述。     

Formation Potential Field for Trajectory Tracking Control of Multi-Agents in Constrained Space

In this paper, a novel Formation Potential Field method is proposed for the multi-agent formation control. The objective is to control a group of agents to automatically generate and maintain a specific formation while avoiding internal collisions and collisions with spatial constraints. A Formation Potential Field is designed combining multiple local attractive potential fields with multiple local repulsive potential fields. To further relax requirements of agents' initial positions and enhance the robustness, a global attractive potential field is added outside the influence range of the local Formation Potential Field. The optimality of the proposed scheme in formation time is analysed as well as illustrated by contrast simulation results. Following the design of the Formation Potential Field, two controllers are proposed accordingly to achieve a stable dynamic formation during the process of trajectory tracking, while the saturation effect of input is taken into account. Furthermore, a collision avoidance strategy based on artificial potential field and Dirac delta function is applied to locally modify the original trajectory of the virtual leader such that agents can avoid collisions with unexpected spatial constraints while maintaining the given formation. Simulation results are presented to illustrate the performance of proposed approaches.     

面向搜索时间受限的完全信息博弈UCT算法改进研究

针对完全信息博弈中搜索时间受限的算法设计问题,在考虑博弈模型不同特点及对结局影响程度的基础上,提出了分阶段的算法模型,给出了三阶段博弈算法设计方法.通过改造影响搜索策略的目标函数,使得在时间受限的前提下,能够方便控制每一阶段均更有效地搜索出较好策略,并给出相应的算法实现与分析.以点格棋为对象,给出了通过改造UCT算法中...     

Constrained least-squares

The method of least-squares is intended to fit a function to a set of data points closely, so as to satisfy a particular criterion of closeness. The approximating function can be taken to be a linear combination of linearly independent functions of an independent variable.If the number of data points is the same as the number of linearly independent primitive functions, then the compound function formed by the linear combination of the primitive functions will fit the data exactly.The method of least-squares is designed to fit an approximant to the data when the order of the approximant is incapable of yielding an exact fit to all the data. For example, given three data points, a linear approximant cannot fit the data points, usually. Likewise, given five data points (or more) a cubic approximant cannot fit the data exactly.In the classical method, all data points are considered to be equally important (or equally doubtful). If some of the data points are more reliable than others, it is easy to attach weights to these points, and with this modification to proceed with the classical method.But sometimes, some of the data points are absolutely reliable. This might suggest that we attach infinite weights to such points, somehow. But this presents arithmetical difficulties.We call “absolutely reliable” data points, “hard points”, and all the other points “soft points”, a terminology borrowed from the automobile, aircraft, and ship-building industries.The following paper shows how we can generate a compound approximant that fits the hard points exactly, and satisfies the least-squares criterion for all the soft points. Then the classical method exhibits itself as a special case of this more general solution.     

Constrained poststratification

Adjustment for covariates (or poststratification) is frequently used in the analysis of randomized clinical trials. The purpose of such analysis is mainly to eliminate some residual bias resulting from any imbalance between treatment groups for some important covariates. Usually, covariate effect is modeled with the data at hand. In this paper, we present a new method of poststratification ("constrained poststratification") which consists of estimating the prognostic significance of covariates in a large historical data base, transferring the model's coefficients into the (smaller) randomized trial data set, and estimating treatment effects conditional on this a priori information. In a simulated experiment, constrained poststratification allowed not only reduction of the bias but also enhancement of the efficiency of the estimation of treatment effect.     

Decentralized Three Dimensional Formation Building Algorithms for a Team of Nonholonomic Mobile Agents

International Journal of Control, Automation and Systems - This article studies 3D formation building in three dimensional spaces by a team of mobile robotic sensors. The multi-agent system...     

On the Equations of Kinematics and Dynamics of Constrained Mechanical Systems

The method of constructing of kinematical and dynamicalequations of mechanical systems, applied to numerical realization, isproposed in this paper. The corresponding difference equations, whichare obtained, give a guarantee of computations with given precision. Theequations of programmed constraints and those of constraintperturbations are defined. The stability of the programmed manifold fornumerical solutions of the kinematical and dynamical equations isobtained by means of corresponding construction of the constraintperturbation equations. The dynamical equations of system withprogrammed constraints are set up in the form of Lagrange equations ingeneralized coordinates. Certain inverse problems of rigid body dynamicsare considered.     

Constrained max-min controllability

The max-min controllability concept of Heymann, Pachter, and Stern [1] is examined for a more general class of problems. Both strong and weak max-min controllability are characterized. It is shown that when magnitude constraints on the control are present, the two max-min controllability concepts are not the same. This differs from the result for problems without control magnitude constraints where it has been shown that the two controllability concepts are identical.     

Constrained delaunay triangulations

Given a set ofn vertices in the plane together with a set of noncrossing, straight-line edges, theconstrained Delaunay triangulation (CDT) is the triangulation of the vertices with the following properties: (1) the prespecified edges are included in the triangulation, and (2) it is as close as possible to the Delaunay triangulation. We show that the CDT can be built in optimalO(n logn) time using a divide-and-conquer technique. This matches the time required to build an arbitrary (unconstrained) Delaunay triangulation and the time required to build an arbitrary constrained (non-Delaunay) triagulation. CDTs, because of their relationship with Delaunay triangulations, have a number of properties that make them useful for the finite-element method. Applications also include motion planning in the presence of polygonal obstacles and constrained Euclidean minimum spanning trees, spanning trees subject to the restriction that some edges are prespecified.An earlier version of the results presented here appeared in theProceedings of the Third Annual Symposium on Computational Geometry (1987).     

Constrained approximate controllability

In the paper, constrained approximate controllability for linear dynamical systems described by abstract differential equations with unbounded control operator is considered. Using methods of spectral analysis for linear self-adjoint operators and general constrained controllability results given by Son (1990), necessary and sufficient conditions of the constrained approximate controllability for the piecewise polynomial controls with values in a given cone are formulated and proved. Moreover, as illustrative examples, constrained approximate boundary controllability of one-dimensional distributed parameter dynamical systems described by partial differential equations of parabolic type with Dirichlet and Neumann boundary conditions are investigated. The constrained controllability conditions obtained in the paper represent an extension of the unconstrained controllability results given by Glothin (1974) and Klamka (1991)     

Constrained Graph Layout

Most current graph layout technology does not lend itself to interactive applications such as animation or advanced user interfaces. We introduce the constrained graph layout model which is better suited for interactive applications. In this model, input to the layout module includes suggested positions for nodes and constraints over the node positions in the graph to be laid out. We describe four implementations of layout modules which are based on the constrained graph layout model. The first three implementations are for undirected graph layout while the fourth is for tree layout. The implementations use active set techniques to solve the layout. Our empirical evaluation shows that they are quite fast and give reasonable layout.     

Constrained delaunay triangulations

Given a set ofn vertices in the plane together with a set of noncrossing, straight-line edges, theconstrained Delaunay triangulation (CDT) is the triangulation of the vertices with the following properties: (1) the prespecified edges are included in the triangulation, and (2) it is as close as possible to the Delaunay triangulation. We show that the CDT can be built in optimalO(n logn) time using a divide-and-conquer technique. This matches the time required to build an arbitrary (unconstrained) Delaunay triangulation and the time required to build an arbitrary constrained (non-Delaunay) triagulation. CDTs, because of their relationship with Delaunay triangulations, have a number of properties that make them useful for the finite-element method. Applications also include motion planning in the presence of polygonal obstacles and constrained Euclidean minimum spanning trees, spanning trees subject to the restriction that some edges are prespecified.     

Constrained regularized differentiation

Numerical differentiation is an ill-posed problem. This article demonstrates that the application of the regularization theory together with the methods of projections on convex sets of constraints improve the accuracy of the derivatives calculation. Simulation results are presented and the applications of the proposed method to edge detection are discussed     

Constrained Maximum Likelihood

Constrained Maximum Likelihood (CML), developed at Aptech Systems, generates maximum likelihood estimates with general parametric constraints (linear or nonlinear, equality or inequality), using the sequential quadratic programming method. CML computes two classes of confidence intervals, by inversion of the Wald and likelihood ratio statistics, and by simulation. The inversion techniques can produce misleading test sizes, but Monte Carlo evidence suggests this problem can be corrected under certain circumstances.     

On Stability of Constrained Receding Horizon Control with Finite Terminal Weighting Matrix

In this paper, a new stabilizing receding horizon control, based on a finite input and state horizon cost with a finite terminal weighting matrix, is proposed for time-varying discrete linear systems with constraints. We propose matrix inequality conditions on the terminal weighting matrix under which closed-loop stability is guaranteed for both cases of unconstrained and constrained systems with input and state constraints. We show that such a terminal weighting matrix can be obtained by solving a linear matrix inequality (LMI). In the case of constrained time-invariant systems, an artificial invariant ellipsoid constraint is introduced in order to relax the conventional terminal equality constraint and to handle constraints. Using the invariant ellipsoid constraints, a feasibility condition of the optimization problem is presented and a region of attraction is characterized for constrained systems with the proposed receding horizon control.