Experimental investigation of stochastic parafoil guidance using a graphics processing unit

Control of autonomous systems subject to stochastic uncertainty is a challenging task. In guided airdrop applications, random wind disturbances play a crucial role in determining landing accuracy and terrain avoidance. This paper describes a stochastic parafoil guidance system which couples uncertainty propagation with optimal control to protect against wind and parameter uncertainty in the presence of impact area obstacles. The algorithm uses real-time Monte Carlo simulation performed on a graphics processing unit (GPU) to evaluate robustness of candidate trajectories in terms of delivery accuracy, obstacle avoidance, and other considerations. Building upon prior theoretical developments, this paper explores performance of the stochastic guidance law compared to standard deterministic guidance schemes, particularly with respect to obstacle avoidance. Flight test results are presented comparing the proposed stochastic guidance algorithm with a standard deterministic one. Through a comprehensive set of simulation results, key implementation aspects of the stochastic algorithm are explored including tradeoffs between the number of candidate trajectories considered, algorithm runtime, and overall guidance performance. Overall, simulation and flight test results demonstrate that the stochastic guidance scheme provides a more robust approach to obstacle avoidance while largely maintaining delivery accuracy.     

Software for estimation of second order effective material properties of porous samples with geometrical and physical nonlinearity accounted for

A method and an algorithm for numerical estimation of effective mechanical properties of porous materials are presented. The effective properties are sought in the form of the nonlinear relation between the second Piola–Kirchhoff stress tensor and the Green strain tensor for anisotropic materials with second-order nonlinearities accounted for. The effective characteristics of test models are computed by means of a CAE Fidesys program module based on the proposed algorithm. The effective material properties as functions of porosity are examined. The finite element mesh that contained more than a million of elements was used while performing stress analysis of a specimen. To reduce computing time, assembly and solution of the global equation system was done in parallel using CUDA technology. The computations were carried out on NVIDIA Tesla C2050 graphics processors. Our results show that accounting for nonlinear effects is essential for correct estimation of effective properties of porous materials.     

一种用于大数据分析服务的启发式云爆发算法

基于异构云联合的并行化大数据分析服务可以提升性能。然而由于大数据网络传输存在较大时延,原则上必须在并行化水平和大数据分析性能之间进行折衷。鉴于此,提出一种启发式云爆发算法用于并行化大数据分析服务。首先确定联合云中哪些计算结点应该用于大数据分析并行处理,然后将大数据妥善地分配给这些计算结点,确保处理同步完成且性能最优,最后,确定被分配的不同大小数据块在各个结点的计算次序,确保数据块传输尽量在结点上一数据块计算期间完成。与其他负载均衡算法做了对比,结果表明,使用该算法后性能可提升20%~60%。     

SIMD向量指令的非满载使用方法研究

大规模SIMD体系结构提供了更强的向量并行硬件支持,但是,大量迭代次数不足的循环由于不能提供足够的并行性,难以用等价的向量方式实现。为了更有效地利用SIMD,提出了一种非满载地使用SIMD指令的向量化方法。研究了向量寄存器的使用方式,基于非满载的向量寄存器使用方式实现了非满载的向量操作和短循环的向量化,并将非满载的向量化方法用于一般循环的向量化。提供了收益分析方法来为本向量化方法作精确指导。实验结果表明了该方法的有效性,所选测试用例的目标循环被向量化,平均加速比达到1.2。     

归约算法统一描述

归约算法在并行计算中应用广泛,目前有很多归约算法应用于不同的情形。这些归约算法各不相同, 逻辑拓扑是 造成区别的关键 。为了统一描述归约算法,揭示它们的共性,给出了一个逻辑拓扑的定义及其性质。在此基础上,给出了归约算法的统一描述,以利于对归约算法的理解,从而设计适应不同应用和环境的归约算法。该描述也可视为可集成不同语义的归约算法框架,从而有助于设计具有新语义的归约算法。本质上,该统一描述是一个归约算法形式定义,有助于验证归约算法的正确性。     

有限元网格积分算法在MIC众核平台上的并行实现

基于英特尔集成众核(Many Integrated Core,MIC)架构,将有限元网格积分算法在至强融核(Xeon Phi)协处理器做了移植和性能分析。该应用全面测试了有限元分析的核心计算过程在MIC上的加速效果,实现了卸载模式(offload)[1]下利用OpenMP在MIC上的线程并行化。计算性能测试结果显示集成众核平台可以有效地加速有限元网格积分算法:1)一块被充分利用的MIC设备卡(3115A)的计算能力超过两路16核Intel XeonTM E5-2670 CPU;2)MIC并发的物理线程可能由于公共缓存访问存在竞争而降低程序的扩展性。测试结果还显示了在多CPU多MIC平台上进一步移植完整的MPI并行有限元模拟软件的可行性。这项工作有助于推动与有限元网格相关的科学和工程高性能计算的研究。     

中国微米纳米-----梳齿式加速度传感器敏感结构等效并联电阻对输出的影响分析

梳齿式加速度传感器敏感结构间感应电容的测量会受到结构间等效电路参数的影响。多层瓷片电容的结构与梳齿式加速度传感器敏感结构类似。本文类比多层瓷片电容中等效电路参数的研究,分析了梳齿式加速度传感器敏感结构间等效电路参数对感应电容测试结果产生的影响。     

Development of a robot assisted carotid blood flow measurement system

The ultrasound diagnosis of the carotid artery is one of the most common non-invasive methods to detect early stage heart and cerebrovascular diseases. However, the precision and repetitiveness of the probe positioning depend exclusively on the operator's skills and dexterity. For this purpose, we propose the development of a robot assisted system to enhance the accuracy and repetitiveness of the probe positioning to measure the wave intensity (WI) index. In this paper, the Waseda-Tokyo Women's Medical-Aloka Blood Flow Measurement Robot System No. 2 Refined (WTA-2R) is presented. The WTA-2R consists of a conventional ultrasound diagnosis system, a 6-DOFs parallel link slave manipulator, a 6-DOFs serial link passive arm, and a master device. Experiments were carried out to verify its effectiveness in terms of accuracy and required time to perform the task. From the experimental results, the positioning accuracy and reduction of required time were confirmed.     

Overactuation: A solution for the accuracy-throughput speed contradiction in parallel axis positioning systems

In parallel axis positioning systems an accuracy-throughput speed contradiction is present. The configuration is often such that the minimum number of actuators from kinematic point of view is used. The mechanical system should possess sufficient stiffness but also low mass. Structural stiffness, and hence accuracy, is obtained at the cost of mass. In intermittent motion systems the moving mass should be minimized, because it limits the attainable acceleration and thus the throughput speed. This dynamic performance barrier can be shifted with additional parallel actuators. To enhance prudence in the initial design phase, knowledge about the opportunities and limitations of parallel axis systems is required. The basic dynamical aspects of parallel axis positioning systems, with a minimum and additional number of parallel actuators, are examined for beam and plate systems. The results of the numerical models are verified with an experimental plate system.     

基于FPGA的CRC并行算法研究与实现

循环冗余校验(CRC)算法广泛应用于通信领域以提高数据传输的可靠性.针对通信过程中常用的CRC校验,介绍了CRC的编码和解码原理,分析了CRC的经典算法的实现过程,并在此基础上提出了基于FPGA的CRC并行处理算法.采用VHDL语言对算法完成建模与实现,并以Altera公司开发的EDA工具QuartusII8.0作为编译、仿真平台进行了仿真验证.电路的综合结果表明,该方法具有更少的资源占用量和更高的工作效率.