Sampling Minimal Subsets with Large Spans for Robust Estimation

When sampling minimal subsets for robust parameter estimation, it is commonly known that obtaining an all-inlier minimal subset is not sufficient; the points therein should also have a large spatial extent. This paper investigates a theoretical basis behind this principle, based on a little known result which expresses the least squares regression as a weighted linear combination of all possible minimal subset estimates. It turns out that the weight of a minimal subset estimate is directly related to the span of the associated points. We then derive an analogous result for total least squares which, unlike ordinary least squares, corrects for errors in both dependent and independent variables. We establish the relevance of our result to computer vision by relating total least squares to geometric estimation techniques. As practical contributions, we elaborate why naive distance-based sampling fails as a strategy to maximise the span of all-inlier minimal subsets produced. In addition we propose a novel method which, unlike previous methods, can consciously target all-inlier minimal subsets with large spans.     

一种掠海无人飞行器SINS/RA/GNSS高度融合估计

研究了掠海无人飞行器SINS/RA/GNSS组合导航的高度融合估计问题。对SINS为公共参考系统的无重置式联邦融合估计方案,提出以SINS天向加速度为量测基准的高度滤波方法,并采用最小二乘估计技术解决导航设备数据更新频率不同导致的量测异步融合问题。仿真结果证明了设计方案的可行性,能够有效提高系统的精度和容错性。     

Vision Based UAV Attitude Estimation: Progress and Insights

Unmanned aerial vehicles (UAVs) are increasingly replacing manned systems in situations that are dangerous, remote, or difficult for manned aircraft to access. Its control tasks are empowered by computer vision technology. Visual sensors are robustly used for stabilization as primary or at least secondary sensors. Hence, UAV stabilization by attitude estimation from visual sensors is a very active research area. Vision based techniques are proving their effectiveness and robustness in handling this problem. In this work a comprehensive review of UAV vision based attitude estimation approaches is covered, starting from horizon based methods and passing by vanishing points, optical flow, and stereoscopic based techniques. A novel segmentation approach for UAV attitude estimation based on polarization is proposed. Our future insightes for attitude estimation from uncalibrated catadioptric sensors are also discussed.     

Robust Gyro‐free Attitude Estimation for a Small Fixed‐wing Unmanned Aerial Vehicle

This paper proposes a backup attitude estimation scheme for small fixed‐wing unmanned aerial vehicles (UAVs) in the event of gyroscopic failure. The attitude is propagated in terms of 3 degrees‐of‐freedom (DoF) aircraft dynamics. The errors in attitude propagation are updated using indirect attitude information obtained from accelerations as sensed by onboard accelerometers and a global positioning system (GPS) receiver. In the event of gyroscopic failure, large uncertainties are introduced into the attitude propagation model. Such uncertainties in states and parameters are modeled as norm‐bound uncertainties and a discrete‐time robust extended Kalman filter (REKF) is implemented to estimate the attitude of the UAV.     

单天线GPS/陀螺仪组合测姿方法研究

针对低成本惯性测量系统的精度容易受引擎震动、陀螺仪漂移的影响,提出了一种适用于活塞引擎的小型UAV姿态测量方法;此方法整合陀螺仪与单天线GPS进行姿态测量,采用以四元数为基础的扩展卡尔曼滤波(EKF)来进行传感器信息融合;利用陀螺仪测得的角速度更新四元数,使用GPS信息所计算的伪姿态来更新滤波器的测量值;仿真结果表明所提出的方法即使在陀螺仪漂移和伪姿态包含噪声的情况下,也拥有较好的长期和短期精度,提升了姿态测量的精度与可靠度。     

一种鲁棒的实时室外光照估计算法

针对动态变化的室外光照求解问题,提出一种室外场景的实时光照估计算法.该算法基于室外场景图像的线性分解模型,利用太阳光基图像中阴影像素的特点和视频中光照的时空连续性构建由数据项和光滑项组成的目标函数,对太阳光和天空光的入射光强进行求解,得到视频帧图像所对应的实时光照参数与相应的太阳光基图像.实验结果表明,文中算法可应用于计算机视觉中受光照变化影响的在线视频处理,如增强现实、视频阴影检测、重光照等.     

姿态传感器采集测试系统的设计与实现

设计了一套基于STM32单片机的姿态传感器无线采集测试系统,大大方便了姿态传感器的研究与应用。首先设计了姿态传感器采集测试系统的整体框架;接着对测试系统的各部分硬件电路进行了说明,描述了ADXL204加速度和ADXRS150陀螺仪传感器的电路和计算公式,并说明了STM32单片机和PC上位机程序的结构和流程;最后利用本测试系统对ADXL204和ADXRS150传感器进行了测试和分析;通过对这两种传感器数据进行卡尔曼滤波,消除陀螺仪的漂移,同时减少了加速度传感器有害的噪声,从而得到精确的角度。     

基于MUSIC/MNM谱估计的鲁棒语音特征提取

针对语音识别系统受噪声干扰识别率急剧下降的问题,通过分析传统的鲁棒语音特征提取方法在语音信号谱估计方面的不足,提出一种在不同信噪比下都具有较好鲁棒性和识别性能的语音特征提取算法．该算法结合多信号分类法（MUSIC）和最小模法（minimum-norm method,MNM）来进行谱估计．接着在移动机器人平台上进行验证实验,结果表明：该算法能有效的提高语音识别率,增强语音识别鲁棒性能．     

一种快速鲁棒运动估计方法

经典的相位法易受高频噪声影响,且计算量大,难以满足实时性要求。提出了基于提升小波变换和离线预建立相关归一化系数与旋转角度对应关系数据库的快速鲁棒运动估计方法。试验结果表明,该方法在保持较高估计精度的情况下,具有较快的估计时间和较好的鲁棒性。     

Robust Navigation in an Unknown Environment With Minimal Sensing and Representation

This paper presents muNav, a novel approach to navigation which, with minimal requirements in terms of onboard sensory, memory, and computational power, exhibits way-finding behaviors in very complex environments. The algorithm is intrinsically robust, since it does not require any internal geometrical representation or self-localization capabilities. Experimental results, performed with both simulated and real robots, validate the proposed theoretical approach.     

基于共轭梯度法和互补滤波相结合的姿态解算算法

为了提高姿态解算精度,提出了一种基于共轭梯度和互补滤波相结合的多传感器数据融合策略。系统采用四元数方法进行姿态解算。利用加速度计和磁强计的输出数据,通过共轭梯度方法对姿态四元数进行寻优估计,再将其和利用陀螺仪输出数据更新的四元数进行互补滤波,解算出姿态角。实验测试表明,这种融合策略使姿态检测系统静态性能和和动态性能均有所提高,尤其在姿态剧烈变化时,其性能明显优于卡尔曼滤波和梯度下降法。     

粒子滤波器及其在卫星姿态估计中的应用

粒子退化和计算量较大是限制粒子滤波应用的主要问题,常规的重采样方法虽然可以缓解粒子退化,但却容易导致粒子枯竭,且计算量较大,因此本文提出了基于混沌摄动的均值逼近粒子滤波器。按权值大小将粒子分组后,用均值替换权值较小的粒子,可使粒子从低似然区向高似然区域逼近。用Kullback信息描述均值逼近产生的粒子分布与似然分布的差别,通过迭代发现Kullback信息是递减的,从而证明该算法是合理的。混沌摄动重采样算法,用类似载波的方法将具有全局遍历性的混沌变量引入,更增加了粒子的多样性。另外,将本算法应用于某型导弹的姿态估计问题中,仿真结果显示了新算法的有效性。     

Fuzzy Relation Equations (II): The Branch-point-solutions and the Categorized Minimal Solutions

This paper presents some novel theoretical results as well as practical algorithms and computational procedures on fuzzy relation equations (FRE). These results refine and improve what has already been reported in a significant manner. In the previous paper, the authors have already proved that the problem of solving the system of fuzzy relation equations is an NP-hard problem. Therefore, it is practically impossible to determine all minimal solutions for a large system if P ≠ NP. In this paper, an existence theorem is proven: there exists a special branch-point-solution that is greater than all minimal solutions and less than the maximum solution. Such branch-point-solution can be calculated based on the solution-base-matrix. Furthermore, a procedure for determining all branch-point-solutions is designed. We also provide efficient algorithms which is capable of determining as well as searching for certain types of minimal solutions. We have thus obtained: (1) a fast algorithm to determine whether a solution is a minimal solution, (2) the algorithm to search for the minimal solutions that has at least a minimum value at a component in the solution vector, and (3) the procedure of determining if a system of fuzzy relation equations has the unique minimal solution. Other properties are also investigated.     

Simultaneous Target Flying Mode Identification and Altitude Estimation in Bistatic T/R‐R HFSWR

High frequency surface wave radar (HFSWR) cannot detect target altitude directly and is unable to identify flight mode. Assuming that the estimation and identification processes are mutually dependent, an integrated method called simultaneous identification and estimation (SIE) is proposed by applying a two‐level multiple model approach to the flight mode probability mass function (pmf) and state probability density function (pdf) simultaneously. The SIE involves two altitude estimation mode‐conditioned model sets, which are established utilizing the variation property of the propagation attenuation at different flight modes. The multiple model approach incorporated in SIE is different from the traditional interacting multiple model (IMM). It is applied at two levels: within each mode‐conditioned estimation model set and across all the mode‐conditioned estimation model sets. Simulations and real trials demonstrate the performance of the proposed SIE method.     

基于YOLOv3和EPnP算法的多药盒姿态估计

针对机械臂药盒抓取操作中对药盒定位和姿态估计的要求,提出一种基于YOLOv3深度学习算法和EPnP算法相结合的多药盒姿态估计方法,此方法主要分为多药盒定位和姿态估计两部分;首先通过YOLOv3算法实现药盒的快速精确定位,并通过定位框分割出单个药盒;然后进行特征提取和特征匹配并估计单应矩阵;通过单应矩阵的透视矩阵变换求得药盒平面4个角点的像素坐标并作为EPnP求解所需的2D点,结合药盒先验尺寸信息在相机坐标系下构建药盒对应的3D点坐标以实现药盒姿态求解;通过结合OptiTrack系统设计了药盒姿态精度对比实验,结果表明,该算法充分发挥了YOLOv3算法兼具快速性和准确性的优势,并且具有良好的姿态估计精度,总体算法速度达到15 FPS,药盒姿态估计平均误差小于0.5°。     

Performance Comparison of Gyro-Based and Gyroless Attitude Estimation for CubeSats

In this paper, a comparison study between gyro-based and gyroless approaches for spacecraft attitude estimation is presented. Due to its vulnerability to the model errors, the gyroless approach has not been widely focused on and there are only few comparison studies available. However, this conventional wisdom might not directly apply to CubeSat attitude estimation, where noisy MEMS gyro is usually implemented. Although the noise density can be improved by low-pass filtering, it sacrifices the bandwidth so that it can induce a discretization error when spacecraft rotates in high speed. This paper outlines expected pros and cons of gyroless attitude estimation with respect to cost and miniaturization, rotational agility, and model errors. Additionally, linearized system models for both of the attitude estimation methods are formulated and a simple guideline for tuning process noise against the model errors is proposed. Numerical results for a realistic earth observation scenario are presented to quantitatively compare the benefits and drawbacks of each attitude estimation method.

     

基于Mahony滤波器和PID控制器的四旋翼飞行器姿态控制

四旋翼飞行器由于其简单的气动布局和复杂的动力学模型在控制领域引起了研究热潮,姿态估计与控制器设计一直是实现四旋翼飞行器稳定飞行的难点。为实现精确的四旋翼飞行器姿态估计,首先分析了IMU传感器示值组成和误差存在的原因,然后在方向余弦矩阵（DCM）和重正交化的基础上,具体给出了Mahony滤波器的实现流程。通过与扩展卡尔曼滤波器对比表明,该算法不仅能保证很高的姿态估计精度,而且计算时间小于扩展卡尔曼滤波器,有助于提高系统姿态估计的实时性。结合Mahony滤波后的姿态信息,采用嵌套PI-PID控制策略设计了控制器。最后,将姿态估计算法和控制算法应用到实验平台上,可以实现四旋翼飞行器悬停和角度跟踪功能。     

快速鲁棒的全局运动估计算法

对全局运动估计算法进行研究,提出一种新的快速且鲁棒性较好的全局运动估计算法.全局运动估计是MPEG-4中的sprite编码中的关键技术,其性能将直接影响编码效率和速度.所提出的算法比传统的Levenberg-Marquadet方法要快,而且参数估计比较准确.新算法是基于非线性密度进行估计的,采用了六参数的仿射模型.为了提高计算速度,采用了3层金字塔进行多分辨率计算,而且在每层迭代计算中都抽取重要的特征点进行计算,同时结合了Gauss-Newton优化计算方法.为了保证计算的准确性,采用基于直方图和基于块两