In the postcapture stage, relative motion could exist between the target and the robot end-effector, which is called unfirm capture . Unmodeled and time-varying dynamic and kinematic coupling in unfirm capture cases obstructs the estimation of target inertia properties and relative motion states. To solve this problem, vision navigation information and general dynamic equations are combined to compensate unknown coupling dynamics based on the extended Kalman filter. To avoid disturbing force and torque reducing the precision of the state estimation, a recursive least square multiplicative extended Kalman filter is developed. By minimizing deviations between expected statistics and real statistics of process noise, this novel filter estimates parameters of the external disturbance. Simulations are carried out to demonstrate the efficiency and effectiveness of the proposed filter.

中文翻译：

---

一种用于太空机器人的创新过滤器，用于不稳定地捕捉翻滚的目标

在捕获后阶段，目标与机器人末端执行器之间可能存在相对运动，称为不确定捕获。在不确定的捕获情况下，未建模和随时间变化的动态和运动学耦合阻碍了目标惯性特性和相对运动状态的估计。为了解决这个问题，结合视觉导航信息和一般动力学方程，基于扩展卡尔曼滤波器补偿未知耦合动力学。为了避免干扰力和转矩降低状态估计的精度，开发了一种递归最小二乘乘法扩展卡尔曼滤波器。通过最小化过程噪声的预期统计和实际统计之间的偏差，这种新颖的滤波器可以估计外部干扰的参数。