基于多传感器时域特征的实时人体行为识别(英文)

提出了一种基于加速度计、陀螺仪、气压计输出时域特征的高精度、高实时性的人体行为模式识别算法。该算法选取多传感器输出的时域特征值作为唯一特征量,通过特征提取运算实现行为的实时识别。通过在实验室自主研发的软硬件平台上进行测试,在识别时间缩短到2 s一次的条件下,对于8种人体日常行为模式和4种摔倒模式的平均识别率可达到94%以上。该算法对于现有算法实时精度有明显提高,且拓展了模式识别的种类,在可穿戴智能终端领域具有很好的应用前景。     

基于改进交互多模型概率数据关联的机动目标跟踪（英文）

为了提高杂波条件下的空中机动目标跟踪精度,提出了一个改进的交互多模型概率数据关联算法。该算法将交互多模型、去偏转换测量和概率数据关联算法相结合,利用交互多模型算法模型集合间不同模型的相互切换来估计跟踪目标的状态;利用去偏转换测量算法对转换测量误差进行去偏补偿,从而减小观测数据坐标变换引起的误差;利用概率数据关联算法处理数据关联和测量的不确定性。通过将本文的算法和基于扩展卡尔曼滤波的概率数据关联算法进行对比分析和验证,实验结果表明本文提出的算法可以提高机动目标的跟踪精度,且跟踪精度相对基于扩展卡尔曼滤波的概率数据关联算法减少26.38%的位置误差。     

四旋翼飞行器建模及位置跟踪控制(英文)

相对其他无人飞行器平台,四旋翼飞行器有其独特的优势,因而受到广泛的关注。位置跟踪控制对四旋翼飞行器的应用非常重要。在阐述四旋翼飞行器的飞行原理和操控机制的基础上,研究了其动力学模型,并提出了一种简化的数学模型。四旋翼飞行器是欠驱动耦合系统,为了实现系统解耦并得到清晰的控制回路,设计了多回路PID控制方案,其控制目标是位置和偏航角,而姿态角和横滚角由位置误差调节。最后,通过仿真验证了控制方法的有效性。     

基于可见光与红外图像特征融合的目标跟踪

针对单一图像源下目标跟踪精度不高的问题,利用跟踪状态下的目标存在于可见光与红外图像中的特征对连续自适应均值移动跟踪算法做出改进。首先选取可见光图像的“颜色梯度背投影”作为改进的目标模型,选取红外图像的“灰度梯度背投影”作为改进的目标模型;然后根据可见光序列图像和红外序列图像各自进行连续自适应均值移动跟踪算法得到的对应的qi系数判定两种图像跟踪的效果,对两种图像的权重进行自适应调整,得到这两种图像的特征级融合图像和跟踪结果。实验结果表明,对于320像素×240像素的可见光和红外图像,基于可见光与红外图像特征融合的目标跟踪算法在复杂背景下能够较准确的跟踪目标,目标跟踪精度为0.5像素,跟踪速度为30~32 ms/帧。     

一种基于线特征的RGB-D视觉里程计算法

基于特征点的视觉里程计在点特征稀少的环境下难以得到足够的匹配点对,从而导致相机运动估计失败,因而提出采集人造环境中特征明显的边缘作为线特征来提高视觉里程计算法的稳定性.采用深度相机获取的RGB图像进行LSD线特征提取,推断线特征对应的图像位置的深度信息,避免深度缺失,将线段上的2D点反投影为3D点,拟合3D点为三维直线...     

基于点线特征视觉惯性融合的机器人SLAM算法

针对弱纹理环境下单目视觉SLAM系统只依靠提取点特征鲁棒性较差的问题,提出一种点线特征视觉与惯导融合的机器人SLAM算法。首先,采用自适应加权提取点线特征并使用普吕克坐标法表示线段,减小计算量同时较好克服线特征提取时线段割裂的不足;其次,采用四叉树法实现点线特征提取均匀化解决特征堆积问题,同时消除点线特征误匹配,再利用视觉点线信息与IMU紧耦合优化机制提高机器人SLAM算法精确度。最后,将该算法在EuRoC数据集和弱纹理环境中进行实验,结果表明,改进后线特征提取相较于传统线特征提取鲁棒性提高了12.94%,相较于原生算法ORB-SLAM3,改进后特征匹配时间节约了19.2%,大型弱纹理环境中绝对定位精度提高了55.6%,所提算法在弱纹理环境中定位效果具有较强的鲁棒性和精确性。     

基于深度特征正射匹配的无人机视觉定位方法

在卫星拒止条件下无人机安全、可靠地完成各类作业的基础是获取高精度的定位信息,传统图像匹配方法保障困难、定位精度差且匹配约束多。因此,提出一种基于深度特征正射匹配的视觉定位方法,通过深度学习网络提取正射校正后的无人机航拍图像和商业地图的深度特征,获得匹配关系,进而计算无人机高精度位置信息。根据视觉测量机理模型分析不同因素对视觉定位精度的影响,使用中空航拍图像数据集进行离线实验,实验结果表明：相比传统基于方向梯度直方图（HOG）特征的模板匹配方法,所提方法的定位精度提高了25%,位置均方根误差（RMSE）优于15 m+0.5%H(5000 m以下),具有一定的工程应用价值。     

基于多弹簧模型的空间梁柱单元(Ⅰ):理论模型

引入多轴应力状态下的塑性应力-应变关系理论,在单元模型中考虑了弹塑性区域剪切变形对单元的弹塑性刚度影响,推导了考虑剪切变形弹塑性刚度影响的多弹簧模型的空间粱柱单元刚度矩阵,为整体钢框架结构的弹塑性动力分析奠定了单元基础.     

基于多弹簧模型的空间梁柱单元(I):理论模型

引入多轴应力状态下的塑性应力-应变关系理论,在单元模型中考虑了弹塑性区域剪切变形对单元的弹塑性刚度影响,推导了考虑剪切变形弹塑性刚度影响的多弹簧模型的空间梁柱单元刚度矩阵,为整体钢框架结构的弹塑性动力分析奠定了单元基础。     

基于特征编码和动态路由优化的视觉定位方法

针对视角变化、光照变化、大尺度和动态物体等复杂场景下,移动机器人定位的准确性低、鲁棒性差等问题,提出基于特征编码和动态路由优化的视觉定位方法。首先,引入基于残差网络的特征编码策略,提取图像的几何特征和语义特征,减少图像噪声信息,加快模型的收敛速度;其次,通过熵密度峰值优化网络的动态路由机制,采用向量表示特征之间的空间位置关系,提升图像特征提取和表达能力,优化网络整体性能;最后,融合优化后的特征编码和动态路由网络,将全局特征描述符和特征向量相结合,保留特征间的差异性和关联性,计算图像特征的相似性用于闭环检测。实验结果表明,相比基于VGG、AlexNet、BoVW及GIST的视觉定位方法,所提方法的准确率分别提高了24.54%、23.06%,43.81%和42.69%,实现了复杂场景下移动机器人闭环检测,提高了定位和建图的准确性和鲁棒性。     

Performances of feature tracking in turbulent boundary layer investigation

In this paper, we describe the application of a feature tracking (FT) algorithm for the measurement of velocity statistics in a turbulent boundary layer over a flat plate at Re _θ ≃ 3,700. The feature tracking algorithm is based on an optical flow approach. Displacements are obtained by searching the parameters of the mapping between interrogation windows in the first and second image which minimize a correlation distance between them. The correlation distance is here defined as the minimum of the sum of squared differences of interrogation windows intensities. The linearized equation which governs the minimization problem is solved with an iterative procedure only where the solution is guaranteed to exist, thus maximizing the signal-to-noise ratio. In this process, the interrogation window first undergoes a pure translation, and then a complete affine deformation. Final mapping parameters provide the velocity and velocity gradients values in a lagrangian framework. The interpolation inherent to window-deforming algorithms represents a critical factor for the overall accuracy and particular attention must be devoted to this step. In this paper different schemes are tested, and their effects on algorithm accuracy are first discussed by looking at the distribution of systematic and random errors computed from synthetic images. The same analysis is then performed on the turbulent boundary layer data, where the effects associated with the use of a near-wall logical mask are also investigated. The comparison with single-point data gathered from the literature demonstrate the overall ability of the FT technique to correctly extract all relevant statistical quantities, including the spanwise vorticity distribution. Concerning the mean velocity profile, no evident influence of the interpolation scheme appears, while the near-wall accuracy is improved by the application of the logical mask. On the contrary, for the fluctuating components of the velocity, the interpolation based on B-Spline basis functions is found to perform better compared to the classical Bicubic scheme, particularly in the highly sheared region close to the wall.

---

A quality criterion for visual tracking during nystagmus

The aim of this study is to construct a quantitative characteristic for the degree of gaze stabilization on a visual target with consideration of physiological constraints. A quality criterion for visual tracking during optokinetic vestibulocervical nystagmus is proposed. The index of imaginary tracking during rotation in the dark allows one to speak of a difference in the image stabilization quality between a healthy person and a patient. Experimental nystagmus data are used to calculate the quality index of visual tracking and the index of imaginary tracking.     

Consensus tracking via quantized iterative learning control for singular nonlinear multi-agent systems with state time-delay and initial state error

Nonlinear Dynamics - This paper investigates the quantized iterative learning consensus tracking problem for singular nonlinear multi-agent systems (MASs) in the presence of state time-delay and...     

基于机器学习与步态特征辅助的行人导航方法

针对微惯性测量组件(MIMU)足部安装的穿戴式导航系统,在包含惯性传感器信息超量程的行进中无法有效进行定位与导航的问题,提出了一种基于虚拟惯性测量组件(VIMU)与步态特征辅助修正惯导系统的行人导航方法。以相同频率采集人体腿部与足部的MIMU数据作为训练样本,通过视觉几何组-长短期记忆混合(VGG-LSTM)神经网络模型拟合两个部位MIMU信息之间的非线性映射关系,构建足部VIMU与虚拟惯性导航系统(VINS);基于人体各步态相位中足部姿态具有高度重复性的特征,对足部VINS姿态信息进行误差修正,并结合足部磁传感器信息确定人体运动的航向信息。实验结果表明,结合零速修正(ZUPT)方法,所提出的VINS构建与误差修正方法可有效提高足部MIMU超量程时行人导航系统性能的可靠性,其高过载运动中的定位误差约为总行进距离的2.5%。     

Finite-time consensus tracking of second-order multi-agent systems via nonsingular TSM

The finite-time consensus tracking problem for second-order multi-agent system is investigated in this paper. The multi-agent system here is composed of a leader with bounded input signal and n followers with bounded disturbances. A new continuous nonlinear distributed consensus tracking protocol is constructed via nonsingular terminal sliding mode (TSM) scheme. It is proved that the overall system will reach consensus in finite time via Lyapunov theory when appropriately choosing the parameters under directed connected topology. Finally, simulations are performed, and results show that the method is robust and efficient.     

An image processing method for feature extraction and dynamic tracking of particle clusters in CFBs

A new image processing method based on the high-speed camera is proposed to identify, locate, and track clusters. The instantaneous characteristic parameters of particle clusters in the riser of the circulating fluidized bed (CFB) can be acquired, such as solids holdup, vertical velocity, lateral displacement, aspect ratio and near-circularity. Experiments were carried out with glass bead particles, river sand particles and FCC particles. The time series of images of gas–solid flow in a CFB riser with a 100 mm × 25 mm cross-section and 3.2 m in length were obtained using high-speed cameras. The k-means++ clustering algorithm is utilized to identify the clusters, centroid is applied to locate the clusters, and the cross-correlation algorithm is employed to track the specific clusters and number them to get the instantaneous characteristic parameters. The results illustrate that the shapes of clusters in the center area are closest to circle, moving upwards at a uniform speed, while the clusters in the side-wall area are mostly elongated or long chain-like, moving slowly downwards. In the transition area, the clusters are more complex, moving upwards at a constant speed, and having large lateral displacement. The results show that the image processing method used in this study is successful in acquiring the dynamic and structural parameters of the clusters simultaneously.     

Construction of three-dimensional images of flow structure via particle tracking techniques

Construction of three-dimensional images of flow structure, based on the quantitative velocity field, is assessed for cases where experimental data are obtained using particle tracking technique. The experimental data are in the form of contiguous planes of particle images. These contiguous data planes are assumed to correspond to successive spatial realizations in steady flow, or to phase-referenced realizations in an unsteady flow.Given the particle images on contiguous planes, the in-plane velocity fields are determined. Then, the out-of-plane velocity field is obtained using a spectral interpolation method. Application of this method allows, in principle, construction of the three-dimensional vorticity field and the streamline patterns.A critical assessment is made of the uncertainties arising from the in-plane interpolation of the velocity field obtained from particle tracking and the evaluation of the out-of-plane velocity component. The consequences of such uncertainties on the reconstructed vorticity distributions and streamline patterns are addressed for two basic types of vortex flows: a columnar vortex, for which the streamlines are not closed and are spatially periodic in the streamwise direction; and for a spherical (Hill's) vortex exhibiting closed streamline patterns, and no spatial periodicity.     

State-constrained bipartite tracking of interconnected robotic systems via hierarchical prescribed-performance control

This paper investigates the collaborative design problem aiming to achieve state-constrained bipartite tracking of the interconnected robotic systems (IRSs) with prescribed performance. In practical applications, the physical limitations of the robots are inevitable. Besides, it is difficult to ensure that the target trajectory is known for each robot of the IRSs in advance. Thus, it is important to follow the target trajectory and meanwhile obey the state constraint being generated from the physical limitations and the external environment of the IRSs. To this end, we propose a new hierarchical state-constrained estimator-based control framework with the characteristics of low computation complexity and high task adaptability. With limited accessibility of the target trajectory, we newly present the estimator to observe it at each time interval through the interconnections among the robots. The state constraint is never violated throughout the convergence process by using the presented control algorithm. The theoretical proof and simulation results are presented to validate the feasibility of the control framework.