A strong tracking nonlinear robust filter for eye tracking

Non-intrusive methods for eye tracking are important for many applications of vision-based human computer interaction.However,due to the high nonlinearity of eye motion,how to ensure the robustness of external interference and accuracy of eye tracking pose the primary obstacle to the integration of eye movements into today＇s interfaces.In this paper,we present a strong tracking unscented Kalman filter （ST-UKF） algorithm,aiming to overcome the difficulty in nonlinear eye tracking.In the proposed ST-UKF,the Suboptimal fading factor of strong tracking filtering is introduced to improve robustness and accuracy of eye tracking.Compared with the related Kalman filter for eye tracking,the proposed ST-UKF has potential advantages in robustness and tracking accuracy.The last experimental results show the validity of our method for eye tracking under realistic conditions.

Sampling strong tracking nonlinear unscented Kalman filter and its application in eye tracking

The unscented Kalman filter is a developed well-known method for nonlinear motion estimation and tracking. However, the standard unscented Kalman filter has the inherent drawbacks, such as numerical instability and much more time spent on calculation in practical applications. In this paper, we present a novel sampling strong tracking nonlinear unscented Kalman filter, aiming to overcome the difficulty in nonlinear eye tracking. In the above proposed filter, the simplified unscented transform sampling strategy with n＋ 2 sigma points leads to the computational efficiency, and suboptimal fading factor of strong tracking filtering is introduced to improve robustness and accuracy of eye tracking. Compared with the related unscented Kalman filter for eye tracking, the proposed filter has potential advantages in robustness, convergence speed, and tracking accuracy. The final experimental results show the validity of our method for eye tracking under realistic conditions.

Strong Tracking Particle Filter Based on the Chi-Square Test for Indoor Positioning

In recent years,a number of wireless indoor positioning(WIP),such as Bluetooth,Wi-Fi,and Ultra-Wideband(UWB)technologies,are emerging.However,the indoor environment is complex and changeable.Walls,pillars,and even pedestrians may block wireless signals and produce non-line-of-sight(NLOS)deviations,resulting in decreased positioning accuracy and the inability to provide people with real-time continuous indoor positioning.This work proposed a strong tracking particle filter based on the chi-square test(SPFC)for indoor positioning.SPFC can fuse indoor wireless signals and the information of the inertial sensing unit(IMU)in the smartphone and detect the NLOS deviation through the chi-square test to avoid the influence of the NLOS deviation on the final positioning result.Simulation experiment results show that the proposed SPFC can reduce the positioning error by 15.1%and 12.3% compared with existing fusion positioning systems in the LOS and NLOS environment.

Strong tracking adaptive Kalman filters for underwater vehicle dead reckoning

To improve underwater vehicle dead reckoning, a developed strong tracking adaptive kalman filter is proposed. The filter is improved with an additional adaptive factor and an estimator of measurement noise covariance. Since the magnitude of fading factor is changed adaptively, the tracking ability of the filter is still enhanced in low velocity condition of underwater vehicles. The results of simulation tests prove the presented filter effective.

A Strong Tracking Filtering Approach for Health Estimation of Marine Gas Turbine Engine

Monitoring and evaluating the health parameters of marine gas turbine engine help in developing predictive control techniques and maintenance schedules.Because the health parameters are unmeasurable,researchers estimate them only based on the available measurement parameters.Kalman filter-based approaches are the most commonly used estimation approaches;how-ever,the conventional Kalman filter-based approaches have a poor robustness to the model uncertainty,and their ability to track the mutation condition is influenced by historical data.Therefore,in this paper,an improved Kalman filter-based algorithm called the strong tracking extended Kalman filter(STEKF)approach is proposed to estimate the gas turbine health parameters.The analytical expressions of Jacobian matrixes are deduced by non-equilibrium point analytical linearization to address the problem of the conventional approaches.The proposed approach was used to estimate the health parameters of a two-shaft marine gas turbine engine in the simulation environment and was compared with the extended Kalman filter(EKF)and the unscented Kalman filter(UKF).The results show that the STEKF approach not only has a computation cost similar to that of the EKF approach but also outperforms the EKF approach when the health parameters change abruptly and the noise mean value is not zero.

Fuzzy Adaptive Strong Tracking Cubature Kalman Filter

To solve the problem that the choice of softening factor in conventional adaptive strong tracking filter( STF) greatly relies on the experience and computer simulation,a new concept of softening factor matrix is introduced and a fuzzy adaptive strong tracking cubature Kalman filter( FASTCKF) based on fuzzy logic controller is proposed. This method monitors residual absolute mean and standard deviation of each measurement component with fuzzy logic adaptive controller( FLAC),and adjusts the softening factor matrix dynamically by fuzzy rules,which is capable to modify suboptimal fading factor of STF adaptively and improve the filter's robust adaptive capacity. The simulation results show that the improved filtering performance is superior to the conventional square root cubature Kalman filter( SCKF) and the strong tracking square root cubature Kalman filter( STSCKF).

Adaptive unscented Kalman filter for parameter and state estimation of nonlinear high-speed objects

An adaptive unscented Kalman filter （AUKF） and an augmented state method are employed to estimate the timevarying parameters and states of a kind of nonlinear high-speed objects. A strong tracking filter is employed to improve the tracking ability and robustness of unscented Kalman filter （UKF） when the process noise is inaccuracy, and wavelet transform is used to improve the estimate accuracy by the variance of measurement noise. An augmented square-root framework is utilized to improve the numerical stability and accuracy of UKF. Monte Carlo simulations and applications in the rapid trajectory estimation of hypersonic artillery shells confirm the effectiveness of the proposed method.

Low-cost adaptive square-root cubature Kalman filter forsystems with process model uncertainty

A novel low-cost adaptive square-root cubature Kalmanfilter （LCASCKF） is proposed to enhance the robustness of processmodels while only increasing the computational load slightly.It is well-known that the Kalman filter cannot handle uncertainties ina process model, such as initial state estimation errors, parametermismatch and abrupt state changes. These uncertainties severelyaffect filter performance and may even provoke divergence. Astrong tracking filter （STF）, which utilizes a suboptimal fading factor,is an adaptive approach that is commonly adopted to solvethis problem. However, if the strong tracking SCKF （STSCKF）uses the same method as the extended Kalman filter （EKF） tointroduce the suboptimal fading factor, it greatly increases thecomputational load. To avoid this problem, a low-cost introductorymethod is proposed and a hypothesis testing theory is applied todetect uncertainties. The computational load analysis is performedby counting the total number of floating-point operations and it isfound that the computational load of LCASCKF is close to that ofSCKF. Experimental results prove that the LCASCKF performs aswell as STSCKF, while the increase in computational load is muchlower than STSCKF.

ADAPTIVE MULTIPLE MODEL FILTER USING IMM AND STF

In fault identification, the Strong Tracking Filter (STF) has strong ability to track the change of some parameters by whitening filtering innovation. In this paper, the authors give out a modified STF by searching the fading factor based on the Least Squared Estimation. In hybrid estimation, the well known Interacting Multiple Model (IMM) Technique can model the change of the system modes. So one can design a new adaptive filter — SIMM. In this filter, our modified STF is a parameter adaptive part and IMM is a mode adaptive part. The benefit of the new filter is that the number of models can be reduced considerably. The simulations show that SIMM greatly improves accuracy of velocity and acceleration compared with the standard IMM to track the maneuvering target when 2 model conditional estimators are used in both filters. And the computation burden of SIMM increases only 6% compared with IMM.

基于强跟踪滤波器的水中高频振荡放电参数分析

为探明水中放电高频振荡阶段参数及其变化特性,提出一种基于自适应噪声完备集合经验模态分解(CEEMDAN)和强跟踪滤波器的时变参数辨识方法。通过该方法分解水中放电实验平台采集的电压、电流信号得到不同频率特征的信号分量,对最适应原始波形的信号分量开展Hilbert变换并求得相应的瞬时幅值、频率,进而得到所需的电阻和电感。实验数据离散度分析结果表明,放电进程中参数变化具有随机性,故利用强跟踪滤波器进一步对实验数据进行辨识处理,可有效地降低随机放电造成的离散性,并获得具备普适性的电阻值和电感值。偏离度分析结果表明,辨识电阻与测量数据除在气泡崩塌阶段随机性过大外,前期偏离度集中在23.26%以下,降低了偏离度处于80%~110%内数据点的干扰,电感偏离度集中在2.35%以下。该方法能够有效地应用于水中高频振荡放电过程的时变参数处理研究中。

基于粒子群优化的无人车双惯性测量单元姿态融合方法

为提高无人车系统中微机电惯性测量单元(MEMS IMU)的姿态角解算精度,提出了一种基于粒子群优化(PSO)算法和自适应强跟踪无迹卡尔曼滤波(STAUKF)算法的数据融合方法。首先,对两种不同精度的IMU模块通过STAUKF算法进行滤波,然后,利用构造的两类误差函数,引入PSO算法对两种IMU的后验估计进行融合,最后,在搭建的无人车平台上进行测试。试验结果表明,相较于两种单一IMU解算数据,所提出的方法解算获得的横滚轴与俯仰轴角度均方根误差分别减小了56.67%、58.94%,相较于冗余式双IMU系统直接加权平均所解算的数据分减小了36.55%、52.15%,解算精度更高、鲁棒性更强。

基于IMM-JPDA-ISTUKF的车载毫米波雷达多目标跟踪算法

为提高车载毫米波雷达多目标跟踪精度指标,提升道路车辆行驶安全性,文中在交互多模型无迹卡尔曼滤波(IMM-UKF)和联合概率数据关联(JPDA)融合的算法基础上,针对车辆运动状态突变处UKF鲁棒性差、滤波精度低的问题,提出了一种基于改进强跟踪UKF(ISTUKF)的IMM-JPDA-ISTUKF算法。通过模拟道路场景搭建的仿真环境对算法性能进行了验证,且为证明该算法在实际道路工况下跟踪精度的提升,还进行了雷达道路测试,通过雷达在道路上获取的车辆数据进一步验证了该算法的有效性。结果表明,该算法在目标车辆运动状态发生变化时的距离跟踪精度和速度跟踪精度方面均得到了提高。

强跟踪容积卡尔曼滤波在空空导弹制导中的应用

针对传统滤波算法在处理目标复杂机动时非线性逼近能力不足、跟踪精度下降等问题,提出一种强跟踪容积卡尔曼滤波(STCKF)方法。首先,根据战斗机规避空空导弹的机动特征,建立了蛇形机动和桶滚机动两种目标运动模型;其次,引入强跟踪滤波(STF)以增强容积卡尔曼滤波(CKF)对系统状态突变等不确定因素的能力;然后,将STCKF应用于导弹末制导目标运动参数估计中,并通过与CKF、无迹卡尔曼滤波(UKF)和粒子滤波(PF)的对比仿真分析验证了该方法的有效性。仿真结果表明,STCKF具有较强的鲁棒性和系统自适应能力,尤其在目标机动突变时其跟踪误差相比CKF减小约10%,能够满足空空导弹末制导高精度和快速响应要求。

一种自适应滤波与干扰观测器相结合的大型舰船状态估计算法

为满足航母等大型舰船目标的状态估计要求,提出一种由非线性干扰观测器和强跟踪容积卡尔曼滤波算法融合形成的交互多模型强补偿容积卡尔曼滤波算法。引入非线性干扰观测器,完成由外界不确定因素引起的干扰总量的估计,并对观测器稳定性进行证明。使用估计的干扰值实时修正强跟踪容积卡尔曼滤波的过程参数,最终形成交互多模型强补偿容积卡尔曼滤波算法,完成对目标状态相对准确的估计。研究结果表明:新提出的滤波算法能够较为准确地完成对目标状态的估计,与变结构多模型粒子滤波算法、变结构多模型无迹卡尔曼滤波算法和交互多模型强跟踪容积卡尔曼滤波算法相比,在目标位置和速度估计上具有更高的估计精度。

基于ASTUKF的分布式农业车辆路面参数辨识方法

针对分布式驱动农业车辆在路面参数辨识过程中,因路面环境变化出现的状态模型误差和时变噪声,导致辨识结果发散的问题,提出了基于自适应强跟踪无迹卡尔曼滤波(Adaptive strong tracking unscented Kalman filter,ASTUKF)的辨识方法。与传统内燃机农业车辆相比,分布式驱动可以直接获取驱动轮的状态信息,结合含有峰值附着系数和极限滑转率的μ-s曲线模型,建立了无迹卡尔曼滤波(Unscented Kalman filter,UKF)辨识算法的状态方程和量测方程。同时,将强跟踪滤波(Strong tracking filter,STF)和自适应滤波(Adaptive filter,AF)引入辨识算法,用以提高对多变环境的识别精度和鲁棒性,并采用奇异值分解(Singular value decomposition,SVD)解决了迭代过程中出现的非正定矩阵的问题。仿真试验结果表明,在突变噪声环境工况下,ASTUKF辨识结果可以快速收敛至目标值附近,且不受突变噪声的影响,各驱动轮峰值附着系数估计结果的平均绝对误差(Mean absolute error,MAE)分别为0.0144、0.0267、0.0144、0.0267,极限滑转率估计结果的MAE分别为0.0025、0.0028、0.0025、0.0028。实车试验表明,在已耕地和未耕地的试验路面上,ASTUKF辨识结果的均值95%置信区间能够匹配测量值,整车的附着系数辨识结果为0.4061(未耕地)、0.3991(已耕地),极限滑转率辨识结果为0.1484(未耕地)、0.3600(已耕地),可为分布式电动农业车辆作业参数感知提供理论参考。

基于强跟踪UKF的自适应PHD-SLAM算法

传统概率假设密度同时定位与建图(Probability Hypothesis Density-Simultaneous Localization and Mapping,PHD-SLAM)方法缺乏在线自适应调整能力,容易受到不确定噪声、初始系统参数选择以及线性化近似误差的影响,从而导致粒子退化问题,进而影响机器人位姿和地图特征点的估计精度。针对这一问题,本文提出了一种基于强跟踪和无迹卡尔曼滤波(Unscented Kalman filter,UKF),并融合最新观测数据来产生重要性密度的PHD-SLAM算法(Strong Tracking UKF PHD-SLAM,SUPHD-SLAM)。所提算法在重要性采样阶段将上一时刻的机器人位姿和地图特征点增广为联合向量,为了避免传统PHD-SLAM中扩展卡尔曼滤波(Extended Kalman filter,EKF)引入的线性化误差,利用UKF对粒子进行预测,并通过引入强跟踪滤波中的渐消因子修正UKF预测后不精确的位姿状态协方差,保持量测新息正交,从而抑制不确定噪声和不精确初始系统参数设置对状态估计的影响。随后通过UKF更新每个位姿粒子,引导粒子向高似然区域移动,以获得更准确的位姿的重要性密度,从而避免粒子退化。从重要性密度中采样新的位姿粒子,针对每个位姿粒子使用基于UKF的PHD滤波计算地图特征点,并用单簇(Single-Cluster,SC)策略更新每个位姿粒子的权重。最后,提取权重最大的位姿粒子及其对应的地图作为状态估计。仿真实验表明,SUPHD-SLAM相较于PHD-SLAM 1.0和PHD-SLAM 2.0,保证计算效率的同时,能够有效的提高机器人位姿和地图特征点的估计精度。

高超目标强跟踪CKF自适应交互多模型跟踪算法

高超目标运动状态复杂且具有高机动性,传统的交互多模型(IMM)跟踪精度低、收敛速度慢,基于此,提出了一种基于多重渐消因子的强跟踪容积卡尔曼滤波(CKF)自适应交互多模型(AIMM)跟踪算法。以IMM-CKF算法为基础,通过对CKF算法的结构进行分析,在时间更新和量测更新的协方差矩阵中引入强跟踪算法的渐消因子,在线实时调整滤波增益,减小模型不匹配导致的滤波精度下降;在IMM的模型集中选择Singer模型、“当前”统计模型和Jerk模型,并针对模型扩维导致CKF算法中无法Cholesky分解的问题引入奇异值分解(SVD)算法;对IMM算法中马尔可夫矩阵提出自适应算法,通过模型似然函数值对转移概率进行自适应修正,增强匹配模型所占比例。仿真结果表明:所提算法跟踪收敛速度提高了约37.5%,跟踪精度提高了16.51%。

基于强跟踪的移动机器人CQKF-SLAM方法

针对容积正交卡尔曼滤波(CQKF)在同时定位与地图构建(SLAM)中系统状态驱动模型与观测数据存在突变,以及协方差分解引起系统不稳定,导致移动机器人定位精度降低的问题,提出一种基于多重渐消因子强跟踪的SVDCQKF-SLAM方法。采用奇异值分解(SVD)代替CQKF算法中的乔列斯基分解,抑制状态误差协方差矩阵负定性;引入多重渐消因子强跟踪滤波器调节状态预测协方差矩阵。通过仿真实验,将所提SLAM方法与其它SLAM方法进行对比,其结果表明,该方法能够有效降低SLAM过程中的定位误差,对移动机器人同时定位与地图构建有一定参考价值。

基于自适应强跟踪Kalman滤波的GNSS跟踪环路设计

为提高GNSS接收机跟踪环路在复杂环境下的跟踪性能,提出一种基于自适应强跟踪Kalman滤波(ASTKF)的跟踪环路,在传统跟踪环路的基础上,以鉴相器输出为观测量进行自适应强跟踪Kalman滤波,滤波结果用于计算导航滤波器的观测量,同时将伪码频率和载波多普勒频率反馈到码NCO和载波NCO,在ASTKF中使用基于卡方分布的渐消因子计算方法,提升跟踪环路鲁棒性。半物理仿真实验表明,相比于基于Kalman滤波的跟踪环路和基于强跟踪Kalman滤波(STKF)的跟踪环路,所提出方法在水平方向上的位置误差和速度误差减小20%以上,有效提高了卫星导航接收机的定位性能。

未知激励下基于改进强跟踪卡尔曼滤波的结构响应重构

提出一种基于改进强跟踪卡尔曼滤波的结构外部激励计算和响应重构的方法。首先,利用状态空间模型得出结构的外部激励,并将激励与模态坐标结合组成新的状态向量,构建增秩状态空间模型。然后,在传统强跟踪卡尔曼滤波算法的基础上进行改进,使其能够处理有色噪声。最后,利用结构部分测点的加速度响应,实现对结构外部激励的计算及其余未测点的速度、加速度响应的重构,分别通过二维桁架和外伸梁进行数值模拟和试验分析,用来验证所提方法的有效性。结果表明,该方法能够有效地重构结构外部激励、未测点的速度和加速度响应,其响应时程曲线与计算响应或测量响应时程曲线吻合良好。