基于序列图像的非合作目标自主导航及验证

针对空间自旋或章动运动的非合作目标运动感知与估计问题,给出了一种基于序列图像的运动测量与状态估计方法。所给出的方法由两部分组成:(1)采用立体视觉测量目标三个非共线的特征点,建立目标参考系并实现相对姿态的测量;(2)采用目标运动学和动力学方程,推导了扩展卡尔曼滤波器,实现目标角速度和惯量比的估计。最后,分别给出了数学仿真和物理试验:(1)利用数学仿真验证了所给出的卡尔曼滤波估计算法在不同目标运动情况下的收敛性;(2)利用机械臂模拟目标的自旋运动,利用双目相机采集目标运动图像,试验结果验证了所给出的运动测量与状态估计方法可以连续的给出目标角速度的精确估计。

Non-cooperative Target Autonomous Navigation and Verification Based on Sequence Image

A motion measurement and state estimation method based on sequence image is presented for the motion perception and estimation for non-cooperative target with spatial spin or nutation motion. The method consists of two parts. (1) Stereo vision is used to measure the three non-collinear feature points of the target, to establish the target reference system and to measure the relative attitude. (2) The extended Kalman filter is derived to estimate the target angular velocity and inertia ratio using the target kinematics and dynamics equations. Finally, numerical simulation and physical experiment are conducted. The simulation results verified the convergence of the proposed Kalman filter estimation algorithm under different target motions. The robotic arm is used to simulate the spin motion of the target, and the target moving image is acquired by the binocular camera. The experimental results verify that the proposed motion measurement and state estimation method can continuously provide an accurate estimation of the target angular velocity.