引用本文:蔡安江,刘凯峰,郭师虹,舒展.基于四元数衍生无迹卡尔曼滤波的二段式多旋翼无人机姿态估计算法[J].控制理论与应用,2020,37(2):365~373.[点击复制]
CAI An-jiang,LIU Kai-feng,GUO Shi-hong,SHU Zhan.Quaternion derivative unscented Kalman filter-based two-step attitude estimation algorithm for multi-rotor unmanned aerial vehicle[J].Control Theory and Technology,2020,37(2):365~373.[点击复制]
基于四元数衍生无迹卡尔曼滤波的二段式多旋翼无人机姿态估计算法
Quaternion derivative unscented Kalman filter-based two-step attitude estimation algorithm for multi-rotor unmanned aerial vehicle
摘要点击 2265  全文点击 938  投稿时间：2018-07-18  修订日期：2019-06-04
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2019.80538
  2020,37(2):365-373
中文关键词  多旋翼无人机  四元数  无迹卡尔曼滤波  姿态估计
英文关键词  multi-rotor UAV  quaternion  Unscented Kalman filter  attitude estimation
基金项目  国家自然科学基金
作者单位E-mail
蔡安江 西安建筑科技大学 cai_aj@163.com 
刘凯峰 西安建筑科技大学  
郭师虹* 西安建筑科技大学 632076202@qq.com 
舒展 西安建筑科技大学  
中文摘要
      针对传统多旋翼无人机姿态估计算法难以兼顾高精度、强实时性以及抗干扰能力差的问题，首先基于一种计算量较小的衍生无迹卡尔曼滤波算法（DUKF），在量测更新中，将加速度数据和磁力计数据分为两个阶段进行姿态四元数校正处理，然后从旋转四元数的本质出发，推测出四元数各元素分别包含着不同的姿态角信息，最后将校正四元数分别乘上为降低校正过程中的相互干扰所设计的系数，提出一种基于四元数DUKF的二段式多旋翼无人机姿态估计算法。通过使用PIXHAWK飞控数据，与传统姿态估计算法进行仿真实验对比，实验表明，本文提出算法与传统使用EKF或UKF的姿态估计算法相比，在实时性、解算精度和抗干扰能力方面有较大提升。
英文摘要
      The traditional attitude estimation algorithm for multi-rotor UAV is difficult to balance high-precision, strong real-time and has poor anti-interference ability. To address this problem, a derivative Unscented Kalman filter algorithm (DUKF) with a relatively small computational complexity is used firstly. In the measurement update, the acceleration data and magnetometer data are divided into two phases for attitude quaternion correction processing. Secondly, according to the nature of quaternion, the assumption that each element of the quaternion contains different attitude angle information is made. Finally, the calibration quaternion is multiplied by the coefficient designed to reduce the mutual interference in the calibration process. A quaternion DUKF-based two-step attitude estimation algorithm for multi-rotor UAV is proposed. By using the PIXHAWK flight control data, the simulation results are compared with traditional attitude estimation algorithms. Experiments show that compared with traditional attitude estimation algorithms using EKF or UKF, the algorithm proposed has great improvement in real-time performance, resolution accuracy and anti-interference ability.