基于视觉显著性的两阶段采样突变目标跟踪算法

针对运动突变目标视觉跟踪问题,提出一种基于视觉显著性的两阶段采样跟踪算法.首先,将视觉显著性信息引入到Wang-Landau蒙特卡罗（Wang-Landau Monte Carlo,WLMC）跟踪算法中,设计了结合显著性先验的接受函数,利用子区域的显著性值来引导马尔可夫链的构造,通过增大目标出现区粒子的接受概率,提高采样效率;其次,针对运动序列中平滑与突变运动共存的特点,建立两阶段采样模型.其中第一阶段对目标当前运动类型进行判定,第二阶段则根据判定结果采用相应算法.突变运动采用基于视觉显著性的WLMC算法,平滑运动采用双链马尔可夫链蒙特卡罗（Marko chain Monte Carlo,MCMC）算法,以此完成目标跟踪,提高算法的鲁棒性.该算法既避免了目标在平滑运动时全局采样导致精度下降的缺点,又能在目标发生运动突变时有效捕获目标.实验结果表明,该算法不仅能有效处理运动突变目标的跟踪问题,在典型图像序列上也具有良好的鲁棒性.     

自适应不规则形状扩展目标跟踪算法

针对具有不规则形状的扩展目标跟踪问题,在目标先验形状未知和运动过程中目标形状发生突变的情况下,提出基于星凸形随机超曲面模型的自适应轮廓扩展目标跟踪算法.首先,研究星凸形扩展目标形状的径向函数傅里叶系数自适应设定问题,基于中心轮廓距离提出不规则形状的自适应方法.其次,针对扩展目标在运动过程中可能发生的形状突变问题,结合滑动窗口构造检测统计量对目标形变进行实时检测,并提出自适应轮廓快速逼近算法对形状突变的扩展目标进行实时跟踪.然后,提出一种拟Jaccard距离作为形状估计质量的评价指标.最后,仿真实验验证了本文所提算法的有效性.     

基于迭代平方根容积粒子滤波目标跟踪算法

针对粒子滤波目标跟踪算法粒子退化及跟踪精度问题,提出了一种基于马尔可夫链-蒙特卡罗(MCMC,Markov Chain Monte Carlo)的迭代平方根容积粒子滤波(ISRCPF,iterated square root cubature Kalman particle filter)算法(ISRCPF-MCMC).在该滤波算法中,利用容积数值积分原则计算非线性随机函数的均值和方差,通过正交矩阵分解代替矩阵开方,在生成的粒子滤波建议分布中融入当前量测值,提高对系统后验概率的逼近程度.然后在此基础上融合MCMC抽样算法(MH,Metropolis Hasting)对所选建议分布进行优化,增加粒子多样性,以提高跟踪精度.仿真试验结果表明,ISRCPF-MCMC算法的估计误差与其他算法相比降低至0.403％.     

基于运动补偿和MCMC算法的视频目标跟踪

针对视频目标跟踪领域摄像头运动等问题,提出一种基于二次观测模型的马尔科夫链蒙特卡洛(MCMC)粒子滤波算法。第1次观测通过计算相邻2帧的光流场对运动模型实时修正使其逼近真实的运动方程,第2次观测MCMC粒子滤波步骤。二次观测模型利用图像中的光流信息进行运动补偿实现跟踪。时变的运动模型可以有效提高MCMC方法的效率,减少无效的粒子点数,使其能更快速地收敛到真实值。实验表明对MCMC进行运动补偿可以有效处理摄像头运动问题。     

海上浮动目标图像跟踪研究与仿真

研究浮动目标的图像准确跟踪问题。海上浮动目标容易受到风向、海浪等突变性因素影响,其运动方向和速度都有很强的突变性和非线性,造成图像中前后联系的方向运动参数估计失准。传统的跟踪方法多是基于运动图像的前后联系进行运动跟踪的,由于这种运动突变性的存在,使得运动间的关联性被大幅降低,跟踪结果出现较大偏差,收敛速度降低,造成跟踪结果的不准确。为此提出了一种自适应方向突变因子的浮动目标运动跟踪算法。引入随机分布的运动突变影响算子,在运动估计过程中作为惩罚因子出现,最大程度降低运动中的突变因素影响。仿真结果表明,自适应突变因子算法能够较准确完成海上浮动目标跟踪,避免了传统方法跟踪延迟等问题。     

自适应策略下Heavy-Ball型动量法的最优个体收敛速率

同时使用自适应步长和动量两种优化技巧的AMSGrad在收敛性分析方面存在比自适应步长算法增加一个对数因子的问题.为了解决该问题,文中在非光滑凸情形下,巧妙选取动量和步长参数,证明自适应策略下Heavy-Ball型动量法具有最优的个体收敛速率,说明自适应策略下Heavy-Ball型动量法兼具动量的加速特性和自适应步长对超参数的低依赖性.求解l₁范数约束下的Hinge损失问题,验证理论分析的正确性.     

基于MCMC的无线传感器网络目标定位算法

为解决基于时延和多普勒频率的无线传感器网络运动目标定位问题,提出一种基于马尔科夫链-蒙特卡罗(Markov-chainMonte-Carlo,MCMC)的直接定位算法。基于最大似然准则从各传感器节点接收信号模型中推导目标位置和速度估计的优化函数;针对该优化函数难以得到闭式解的问题,将优化函数转化为马尔科夫链的稳态分布,利用MCMC方法对目标位置和速度参数分布进行抽样,得到目标位置和速度参数的样本,通过统计样本均值得到目标位置和速度的估计值。实例仿真计算结果表明,该算法比现有算法具有更高的定位精度、稳健性和计算效率,在一般信噪比条件下,性能逼近克拉美罗界。     

基于随机梯度的变动量因子自适应白化算法

针对自适应白化技术中算法的收敛速度问题, 通过融入具有变动量因子特性的动量项,提出了一种快速的自适应白化算法. 该算法利用动量项来加速系统的收敛速度,并基于随机梯度方法对动量因子进行自适应更新,有效提升了白化系统的整体性能. 仿真实验表明本文算法在平稳和非平稳环境下具有良好的性能.     

强机动目标跟踪自适应交互式多模型算法

针对机动目标跟踪过程中建立的目标模型和目标的实际运动模式出现失配的问题,提出了从一组离散模型集中选出最优模型,并自适应调整模型参数,使模型逼近目标实际运动模式的交互式多模型算法.蒙特卡罗仿真表明,该算法与传统的常速模型与自适应协同转弯模璎交互算法(IMM-CV/ACT)相比,在目标发生强机动时,能及时有效的把跟踪误差峰值控制在测最标准差之下,适合于强机动目标跟踪.     

基于RSA-FCM算法的运动目标跟踪

运动目标跟踪领域的研究常用颜色直方图作为统计特征, 效果良好但也具有易受光照变化影响等缺点, 运用模糊颜色直方图的跟踪方法能解决以上问题. 针对传统模糊聚类方法中的不足之处, 提出了基于RSA-FCM算法的运动目标跟踪算法, 即在模糊聚类过程中使用随机采样策略确定聚类初值, 同时运用自适应模糊聚类模型进行运算, 提高了跟踪的速度和精度. 实验对比表明, 本文提出的算法在运动目标跟踪准确性和实时性较传统算法都有改进.     

Adaptive Hamiltonian MCMC sampling for robust visual tracking

Recent researches on visual tracking have shown significant improvement in accuracy by handling the large uncertainties induced by appearance variation and abrupt motion. Most studies concentrate on random walk based Markov chain Monte Carlo(MCMC) tracking methods which have shown inefficiency in sampling from complex and high-dimensional distributions. This paper proposes an adaptive Hamiltonian Monte Carlo sampling based tracking method within the Bayesian filtering framework. In order to suppress the random walk behavior in Gibbs sampling stage, the ordered over-relaxation method is used to draw the momentum item for the joint state variable. An adaptive step-size based scheme is used to simulate the Hamiltonian dynamics in order to reduce the simulation error and improve acceptance rate of the proposed samples. Furthermore, in designing the appearnce model, we introduce the locality sensitive histogram (LSH) to deal with appearance changes induced by illumination change. The proposed tracking method is compared with several state-of-the-art trackers using different quantitative measures: success rate and abruption capture rate. Extensive experimental results have shown its superiority to several other trackers.     

Adaptive fusion of particle filtering and spatio-temporal motion energy for human tracking

Object tracking is an active research area nowadays due to its importance in human computer interface, teleconferencing and video surveillance. However, reliable tracking of objects in the presence of occlusions, pose and illumination changes is still a challenging topic. In this paper, we introduce a novel tracking approach that fuses two cues namely colour and spatio-temporal motion energy within a particle filter based framework. We conduct a measure of coherent motion over two image frames, which reveals the spatio-temporal dynamics of the target. At the same time, the importance of both colour and motion energy cues is determined in the stage of reliability evaluation. This determination helps maintain the performance of the tracking system against abrupt appearance changes. Experimental results demonstrate that the proposed method outperforms the other state of the art techniques in the used test datasets.     

基于数据ADP算法的一类带有执行器饱和的未知离散时间系统最优跟踪控制

针对一类带有执行器饱和的未知动态离散时间非线性系统, 提出了一种新的最优跟踪控制方案. 该方案基于迭代自适应动态规划算法, 为了实现最优控制, 首先建立了未知系统动态的数据辨识器. 通过引入M网络, 获得了稳态控制的精确表达式. 为了消除执行器饱和的影响, 提出了一个非二次的性能指标函数. 然后提出了一种迭代自适应动态规划算法获得最优跟踪控制的解, 并给出了收敛性分析. 为了实现最优控制方案, 神经网络被用来构建数据辨识器、计算性能指标函数、近似最优控制策略和求解稳态控制. 仿真结果验证了本文所提出的最优跟踪控制方法的有效性.     

Dynamics and control of a novel 3-DOF parallel manipulator with actuation redundancy

This paper deals with the dynamics and control of a novel 3-degrees-of-freedom （DOF） parallel manipulator with actuation redundancy. According to the kinematics of the redundant manipulator, the inverse dynamic equation is formulated in the task space by using the Lagrangian formalism, and the driving force is optimized by utilizing the minimal 2-norm method. Based on the dynamic model, a synchronized sliding mode control scheme based on contour error is proposed to implement accurate motion tracking control. Additionally, an adaptive method is introduced to approximate the lumped uncertainty of the system and provide a chattering-free control. The simulation results indicate the effectiveness of the proposed approaches and demonstrate the satisfactory tracking performance compared to the conventional controller in the presence of the parameter uncertainties and un-modelled dynamics for the motion control of manipulators.     

A unified symplectic pseudospectral method for motion planning and tracking control of 3D underactuated overhead cranes

In this paper, a unified symplectic pseudospectral method for motion planning and tracking control of 3D underactuated overhead cranes is proposed. A feasible reference trajectory taking constraints into consideration is first generated offline by the symplectic pseudospectral optimal control method. Then, a trajectory tracking model predictive controller also based on the symplectic pseudospectral method is developed to track the reference trajectory. At each sampling instant, the trajectory tracking controller works by solving an open‐loop optimal control problem where linearized system dynamics is used instead to improve the computational efficiency. Since the symplectic pseudospectral optimal control method is the core algorithm for both offline trajectory planning and online trajectory tracking, constraints on state variables and control inputs can be easily imposed and hence theoretically guaranteed in solutions. By selecting proper weighted matrices on tracking error and control, the developed controller could achieve control objectives in both accurate trolley positioning and fast suppressing of residual swing angles. Simulations for 3D overhead crane systems in the presence of perturbations in initial conditions, an abrupt variation of system parameter, and various external disturbances demonstrate that the developed controller is robust and of excellent control performance.     

基于FOE和改进MCMC的视频运动目标跟踪方法

改进的马尔可夫链蒙特卡洛MCMC(Markov Chain Monte Carlo)粒子滤波跟踪算法可以实现稳定跟踪多目标的目的。但在运动场景下,常常出现跟丢或者误跟的情况。考虑到相机聚焦中心FOE(Focus Of Expansion)在估计摄像头运动方面有不可替代的作用,首先通过构建FOE与目标在视频中位置的一个简单估计模型,估计目标的位置,再通过FOE与MCMC的结合,改善了目标丢失和抖动的现象,达到更加准确估计目标的目的。实验表明该方法对摄像头前后平移运动有比较理想的效果。     

Motion/force control scheme for electrically driven cooperative multiple mobile manipulators

This paper presents the motion and force control problem of rigid-link electrically driven cooperative mobile manipulators handling a rigid object. Although, the motion/force control problem of cooperative mobile manipulators has been enthusiastically studied. But there is little research on the motion/force control of electrically driven cooperative mobile manipulators. Due to the inclusion of the actuator dynamics with the manipulator’s dynamics, the controller exhibits some important characteristics. For the electromechanical system, we have designed a novel controller at the dynamic level as well as at the actuator level. In the proposed control scheme, at the dynamic level, uncertain non-linear mechanical dynamics is approximated with a hybrid controller containing model-based control scheme combined with model-free neural network based control scheme together with an adaptive bound. The adaptive bound is used to suppress the effects of external disturbances, friction terms, and reconstruction error of the neural network. At the actuator level, for the approximation of the unknown electrical dynamics, the model-free neural network is utilized. The developed control scheme provides that the position tracking errors, as well as the internal force, converge to the desired levels. Additionally, direct current motors are also controlled in such a way that the desired currents and torques can be attained. In order to make the overall system to be asymptotically stable, online learning of the weights and the parameter adaptation of the parameters is utilized in the Lyapunov function. The superiority of the developed control method is carried out with the numerical simulation results and its superior robustness is shown in a comparative manner.     

Motion priors based on goals hierarchies in pedestrian tracking applications

In this paper, the problem of automated scene understanding by tracking and predicting paths for multiple humans is tackled, with a new methodology using data from a single, fixed camera monitoring the environment. Our main idea is to build goal-oriented prior motion models that could drive both the tracking and path prediction algorithms, based on a coarse-to-fine modeling of the target goal. To implement this idea, we use a dataset of training video sequences with associated ground-truth trajectories and from which we extract hierarchically a set of key locations. These key locations may correspond to exit/entrance zones in the observed scene, or to crossroads where trajectories have often abrupt changes of direction. A simple heuristic allows us to make piecewise associations of the ground-truth trajectories to the key locations, and we use these data to learn one statistical motion model per key location, based on the variations of the trajectories in the training data and on a regularizing prior over the models spatial variations. We illustrate how to use these motion priors within an interacting multiple model scheme for target tracking and path prediction, and we finally evaluate this methodology with experiments on common datasets for tracking algorithms comparison.