确定采样型滤波器优化算法及其在编队导航中的应用

Optimization algorithm of determination sampling type filter and its application in formation navigation

针对基于视觉的编队导航系统的导航参数估计问题,开展了确定采样滤波器对导航参数的估计研究。根据僚机体坐标系下的编队运动方程和视觉量测方程,建立了适用于确定采样滤波器的系统状态方程和量测方程。针对采样点不能满足姿态四元数的归一化要求,采用无约束的罗德里格参数代替四元数进行采样,选择的采样点能够满足编队导航系统的要求。采用3阶采样点的确定采样型滤波器对编队导航状态进行估计。UKF和CKF由于分别存在中心采样点权值过大和非局部效应而导致滤波发散,因此通过正交变换的方法对非局部效应进行修正,实现了估计误差的稳定收敛。通过仿真验证表明,提出的改进方法能够满足编队导航状态估计的要求。

To solve the problem existing in navigation parameter estimation for formation navigation system based on vision, the estimation of navigation parameter by the determine sampling type filter is studied. According to the formation motion equation and vision measuration equation in body frame of wing plane, the systematical state equation and measuration equation suitable for the determination sampling type filter are established. In consideration that the sampling point can not meet the normalization demand of attitude quaternion, the unrestraint Rodrigues parameter is used to replace quaternion. The sampling point selected by it can meet the demand of formation navigation system. The determination sampling type filter with third-order sampling point is used for the estimation of formation navigation state. Since the centre sampling point weight of UKF is oversize, and CKF has non-local effect, which may lead to filter divergence, a method of orthogonal conversion is used to correct non-local effect to realize stable convergence of evaluated error. Simulation results are given to prove the technique.