多矢量姿态测量算法研究综述

A review of algorithms for multi-vector attitude synthesis of research

将矢量测量转化为姿态测量,该问题在姿态估计中被称为Wahba问题。介绍了最小化Wahba问题的损失函数的不同算法,包括最优四元数估计器(ESOQ)、四元数估计器(QUEST)、快速最优姿态矩阵(FOAM)、矩阵的奇异值分解(SVD)等广泛使用的算法,以及基于快速线性四元数姿态估计器(FLAE)、黎曼流形等近年来提出的算法。给出了简单的计算原理与推导过程,并通过计算机仿真对照,归纳各类算法在计算旋转矩阵时的计算精度、鲁棒性。针对姿态测量算法应对当前应用场景的性能需求,简要介绍其在惯性测量单元(IMU)动态对准、协同集群视觉定姿、图像拼接等方面的适用可能和工作原理,并在此基础上简述了姿态测量算法当前的缺陷与未来的发展趋势。

The transformation of vector measurements into attitude measurements is known as the Wahba problem in the literature of attitude estimation.This review introduces various algorithms for minimizing the Wahba’s loss function,including estimator of the optimal quaternion(ESOQ),quaternion estimator(QUEST),fast optimal attitude matrix(FOAM),singular value decomposition(SVD)as well as new algorithms proposed in recent years such as fast linear quaternion attitude estimator(FLAE)and the algorithm based on Riemannian manifold.The calculation principle and derivation process are briefly introduced.The computational accuracy and robustness of various methods in calculating rotation matrices are summarized through computer simulation.Then,this review outlines the applicability and working principles of these algorithms in inertial measurement unit(IMU)dynamic alignment,collaborative cluster visual attitude determination,and image mosaic in addressing emerging performance requirements in current applications.Finally,we describe the current limitations and development trends of attitude algorithms in the future.