空间自旋目标混合EFIR/DFT运动轨迹预测算法

摘要

针对自旋目标运动轨迹的跟踪与预测中鲁棒性与时效性问题,本文提出一种在视觉测量目标位姿的基础上,通过混合扩展有限冲击响应(EFIR)/离散傅立叶变换(DFT)估计目标状态与特征参数,进而预测目标轨迹的方法.在视觉相机对目标特征点位姿测量的基础上,将运动过程分解为平动与转动,时域与频域同步估计目标的状态与动力学参数,采用DFT估计与平动相关参量,采用EFIR估计与转动相关参量,根据空间漂浮目标动力学方程,实现在过程噪声与量测噪声未知的复杂条件下对目标轨迹的长期准确预测,并通过地面机器人模拟试验对预测方法的正确性和有效性开展验证.结果表明:利用本文提出的方法实现了对空间自旋目标运动轨迹的准确预测;与传统基于扩展卡尔曼滤波的预测方法相比,在过程噪声、量测噪声未知的条件下,文中提出的方法有效缩短了参数收敛时间,提高了参数估计与轨迹预测精度.%Based on the visual camera,a robust and efficient method for tracking and predicting the trajectory of a space spinning target is proposed in this paper.The initial discrete point set of a motion trajectory is firstly accumulated in real-time,and then the motion is decomposed into translation and rotation.A hybrid extended finite impulse response (EFIR)/discrete Fourier transform (DFT) is used to estimate the states and dynamics parameters in time domain and frequency domain simultaneously.According to the dynamic equations of a free-floating object,we achieve the long-term and precise prediction while the measurement noise and process noise are unknown.The experiment with ground robot is presented to verify the correctness and effectiveness of the proposed method.The results show that the trajectory of a space spinning target can be predicted accurately using our proposed method.Comparing with the traditional extended Kalman filtering algorithm,the presented control method can improve the the speed of the parameter estimation and the accuracy of the trajectory prediction despite the noise covariance and the initial conditions are not exactly known in advance.