On the dynamic version of the minimum hand jerk criterion

This paper is concerned with the problem of trajectory formation of humanlike reaching movements. First, we review conventional criteria of optimality adopted in robotics and computational neuroscience for the prediction of reaching movements and formulate a dynamic version of the minimum hand jerk criteria. We call it a minimum driving force change criterion and check its performance for the free‐space movements. Next, we test the performance of the new criterion for the movements where the human hand is geometrically constrained by the external environment, and for the movements with flexible objects. The main feature of these movements is that the hand velocity profiles are not always bell shaped. Our simulations and initial experimental results show that the minimum driving force change criterion can roughly capture this feature and, therefore, can be a reasonable candidate for modeling of humanlike reaching movements. © 2005 Wiley Periodicals, Inc.     

Identifying the criterion spontaneously minimized during the take-off phase of a sub-maximal long jump through optimal synthesis

Optimal synthesis of human movement or the prediction of the kinematics of a new movement require not only that the multi-body system be modeled but also that a performance criterion is specified. For sub-maximal movements the selection of a suitable performance criterion, able to generate realistic dynamic behavior is difficult. A two-dimensional simulation model of the take-off phase of a sub-maximal long jump was developed to study the effect of criterion choice on the realism of simulated movements. A parametric optimization technique was employed to obtain solutions to the constrained equations of motion. Seven different criteria were evaluated, by comparing simulated movements with an actual performance, to identify the criterion which most closely approximated that spontaneously minimized by the athlete. Synthesis of the take-off phase of a sub-maximal long jump was found to be sensitive to the chosen criterion, with a criterion based on minimizing joint intersegmental forces found to perform well.     

Quasi optimal sagittal gait of a biped robot with a new structure of knee joint

The design of humanoid robots has been a tricky challenge for several years. Due to the kinematic complexity of human joints, their movements are notoriously difficult to be reproduced by a mechanism. The human knees allow movements including rolling and sliding, and therefore the design of new bio-inspired knees is of utmost importance for the reproduction of anthropomorphic walking in the sagittal plane. In this article, the kinematic characteristics of knees were analyzed and a mechanical solution for reproducing them is proposed. The geometrical, kinematic and dynamic models are built together with an impact model for a biped robot with the new knee kinematic. The walking gait is studied as a problem of parametric optimization under constraints. The trajectories of walking are approximated by mathematical functions for a gait composed of single support phases with impacts. Energy criteria allow comparing the robot provided with the new rolling knee mechanism and a robot equipped with revolute knee joints. The results of the optimizations show that the rolling knee brings a decrease of the sthenic criterion. The comparisons of torques are also observed to show the difference of energy distribution between the actuators. For the same actuator selection, these results prove that the robot with rolling knees can walk longer than the robot with revolute joint knees.     

基于最小代价的多分类器动态集成

本文提出一种基于最小代价准则的分类器动态集成方法．与一般方法不同,动态集成是根据“性能预测特征”,动态地为每一样本选择最适合的一组分类器进行集成．该选择基于使误识代价与时间代价最小化的准则,改变代价函数的定义可以方便地达到识别率与识别速度之间的不同折衷．本文中提出了两种分类器动态集成的方法,并介绍了在联机手写汉字识别中的具体应用．在实验中使了３个分类器进行动态集成,因此,得到７种分类组合．在预先定义的代价意义下,我们比较了动态集成方法和其它７种固定方法的性能．实验结果证明了动态集成方法的高灵活性、实用性和提高系统综合性能的能力．     

Probabilistic tree-based representation for solving minimum cost integer flow problems with nonlinear non-convex cost functions

The minimum cost flow problem (MCFP) is the most generic variation of the network flow problem which aims to transfer a commodity throughout the network to satisfy demands. The problem size (in terms of the number of nodes and arcs) and the shape of the cost function are the most critical factors when considering MCFPs. Existing mathematical programming techniques often assume the cost functions to be linear or convex. Unfortunately, the linearity and convexity assumptions are too restrictive for modelling many real-world scenarios. In addition, many real-world MCFPs are large-scale, with networks having a large number of nodes and arcs. In this paper, we propose a probabilistic tree-based genetic algorithm (PTbGA) for solving large-scale minimum cost integer flow problems with nonlinear non-convex cost functions. We first compare this probabilistic tree-based representation scheme with the priority-based representation scheme, which is the most commonly-used representation for solving MCFPs. We then compare the performance of PTbGA with that of the priority-based genetic algorithm (PrGA), and two state-of-the-art mathematical solvers on a set of MCFP instances. Our experimental results demonstrate the superiority and efficiency of PTbGA in dealing with large-sized MCFPs, as compared to the PrGA method and the mathematical solvers.     

一种最小代价组播树的快速算法

本文对KMB算法进行了改进，提出了一种快速的最小代价组播树算法，它只需使用一次PRIM算法，也不需要判断叶结点，从而快速地获得了最小代价组播树，减少了算法的运行时间。随机网络模型的仿真实验表明：该算法的计算时间远小于KMB算法，是一种快速、稳定、高效的算法。     

基于分级规划策略的拟人机械臂仿人运动规划算法研究

为使拟人机械臂具有高精度的仿人运动,提出一种通过触发条件和分级规划策略 的仿人运动新方法。将人臂运动过程离散为不同运动阶段,在每一个运动阶段都有与之对应的 规划层,在不同的规划层中,拟人机械臂的运动特点不同。利用各自的特点建立不同规划层下 的运动模型及臂姿预测指标,对拟人机械臂臂姿进行预测。最后,以NAO 机器人为实验平台, 比较所提方法与最小势能法(MTPE)的静态臂姿与动态臂姿预测,并与运动捕捉系统(OptiTrack) 采集的真实人臂运动数据进行比较。实验表明,该方法具有较小的静态臂姿和动态臂姿预测误 差,能使拟人机械臂产生高度逼真的仿人运动。     

用于非高斯系统降维的最小残差熵主元网络

基于最小均方误差的主元分析和主元神经网络是有效的多变量降维统计技术,它们所提取的主元含有系统最大方差.非高斯随机系统的近似模型应当含有系统最大信息熵,但包含最大方差并不一定包含最大信息熵.该文提出一种以最小残差熵为通用指标的非线性主元神经网络模型,并给出了一种基于Parzen窗口密度函数估计的熵近似计算方法和网络学习算法.然后从信息论角度分析了,在高斯随机系统中基于最小残差熵和最小均方差为指标的主元网络学习结果具有一致性.最后以仿真验证该方法的有效性,并与基于最小均方误差的主元分析和主元神经网络方法的计算结果进行对比性分析.     

Parameterized energy–latency trade-offs for data propagation in sensor networks

We study the problem of greedy, single path data propagation in wireless sensor networks, aiming mainly to minimize the energy dissipation. In particular, we first mathematically analyze and experimentally evaluate the energy efficiency and latency of three characteristic protocols, each one selecting the next hop node with respect to a different criterion (minimum projection, minimum angle and minimum distance to the destination). Our analytic and simulation findings suggest that any single criterion does not simultaneously satisfy both energy efficiency and low latency. Towards parameterized energy–latency trade-offs we provide as well hybrid combinations of the two criteria (direction and proximity to the sink). Our hybrid protocols achieve significant performance gains and allow fine-tuning of desired performance. Also, they have nice energy balance properties, and can prolong the network lifetime.     

Clustering in Dynamic Spatial Databases

Efficient clustering in dynamic spatial databases is currently an open problem with many potential applications. Most traditional spatial clustering algorithms are inadequate because they do not have an efficient support for incremental clustering.In this paper, we propose DClust, a novel clustering technique for dynamic spatial databases. DClust is able to provide multi-resolution view of the clusters, generate arbitrary shapes clusters in the presence of noise, generate clusters that are insensitive to ordering of input data and support incremental clustering efficiently. DClust utilizes the density criterion that captures arbitrary cluster shapes and sizes to select a number of representative points, and builds the Minimum Spanning Tree (MST) of these representative points, called R-MST. After the initial clustering, a summary of the cluster structure is built. This summary enables quick localization of the effect of data updates on the current set of clusters. Our experimental results show that DClust outperforms existing spatial clustering methods such as DBSCAN, C2P, DENCLUE, Incremental DBSCAN and BIRCH in terms of clustering time and accuracy of clusters found.