动载体上目标物姿态双惯性和视觉组合测量新方法

Attitude measurement on moving base is widely applied in fields such as helmet tracking system, man-machine interactive system of carrier, virtual reality and augmented reality. On the moving carrier, the non-inertial motion of the carrier will act on the measured object on it. And the non-inertial motion can be sensed by the inertial sensor attached to the measured object. Thus, the non-inertial motion of the carrier itself becomes the interference component in the attitude calculation of the measured object, resulting in the failure of vision and single inertial fusion attitude measurement system. Considering the motion interference of the carrier in the attitude measurement of a measured object on the carrier, a novel combined measurement scheme of vision and dual inertial fusion attitude measurement scheme is studied in this subject, aiming at the attitude measurement of the objects on a moving carrier relative to the carrier. The master inertial sensor and the slave inertial sensor are attached to the measured object and the moving carrier to sense the angular motion information of the measured object and the moving carrier respectively. Then the enhanced differential operation is used to separate the carrier motion information in the master inertial sensor with the aid of the slave inertial sensor, preparing for further attitude calculation. The visual attitude measurement is realized by the method of combining the camera with the LED feature points. The main research contents are as follows: the design and execution of the global coordinate system normalization scheme; the research of dual inertial sensor difference operation, multi frequency data fusion algorithm and fault tolerance algorithm of multisource data fusion，improving accuracy and stability of attitude measurement. The measurement experiment system is set up, and the proposed methods and key technologies are verified by experiment.

飞行员头盔瞄准定位、运载体人机交互系统、虚拟现实和增强现实等领域都需进行运动载体上目标物体的姿态测量。载体运动分量作用于与其固连的惯性器件上成为姿态解算中的干扰，使得视觉和单一惯性器件的组合测量失效。本课题研究视觉和双惯性陀螺仪组合的姿态测量新方法，减小或消除载体运动的干扰，实现运动载体上被测物相对载体基准系的姿态测量。主陀螺仪与被测物刚性连接，辅助陀螺仪与动载体刚性连接。它们分别用来感测被测物以及动载体相对空间惯性系的角运动信息，使用增强差分运算借助辅助陀螺仪将主陀螺仪中的载体运动信息分离，以便进行后续姿态解算。视觉姿态测量采用摄像机与LED标识点结合的方法实现。主要研究内容有：设计全局坐标系归一化方案及标定方法；研究双惯性传感器差分算法和多频率数据融合残差补偿算法以及组合测量系统的容错性问题，提高姿态测量准确性和稳定性；组建测量实验系统，对所研究的方法、理论和关键技术进行实验验证。

动载体上目标姿态测量是飞行器、舰船、车辆、运动模拟器等运动载体上进行导航、控制及追踪的关键技术，如飞行员头盔定位、载体上机器人导航与控制、虚拟/增强现实系统等。课题充分利用惯性测量单元IMU大范围快速和视觉测量准确的优点，研究了动载体上目标物姿态双惯性与视觉组合测量方法，优化设计了多传感器组合姿态测量系统，搭建了实验装置，对视觉姿态解算算法、全局坐标系归一化方法、双惯性实时增强差分算法、自适应去噪算法及多频率数据融合算法等进行了实验验证，实现了动载体上目标姿态大范围快速准确测量。主要研究内容和创新性成果有：优化设计了动载体上目标姿态双惯性和视觉组合测量系统，实现了动载体上目标物体相对运动基准坐标系空间姿态的快速准确测量；研究了动载体上目标姿态双惯性和视觉组合测量系统视觉姿态解算算法，提出了一种基于成像光线追踪模型的姿态求解算法，有效提高了视觉姿态解算的准确性和稳定性；研究了多传感器全局坐标系归一化理论和方法，提出了一种双惯性实时增强差分算法，为进一步的数据融合计算奠定基础；研究了IMU信号噪声处理算法，将AMA和基于EMD的去噪算法相结合，提出了一种混合型自适应随机噪声处理算法，有效降低了信号中的随机噪声；研究了多传感器信号数据融合算法，提出了一种鲁棒QCKF数据融合算法，提高了测量系统的鲁棒性及滤波精度，尤其在有异常测量值时刻的滤波精度；提出一种了变分贝叶斯推断自适应误差状态卡尔曼滤波数据融合算法，解决了噪声协方差矩阵无法准确预设的问题；提出了一种多频率数据残差补偿融合算法，建立了状态误差的自更新方程，在低频视觉数据间隔实现了对滤波估计量的修正。