Relative Pose Determination for Noncooperative Spacecraft Under Noninertial Observation Frame

This article addresses the six-degree-of-freedom pose estimation problem for close-range relative navigation of noncooperative spacecraft under non-inertial observation and inertia uncertainties. Based on the unified dual-quaternion description, a compact multiplicative extended Kalman filter framework incorporating rigid-body spacecraft dynamics is developed. The proposed filter enables the determination of relative motion states and identification of inertial parameters, solely relying on noisy pose measurement. An analytical solution for closed-form Jacobian is constructed to deal with the nonlinearity arising from non-inertial observation. By incorporating free-tumbling relative dynamics, the filtering method is designed to identify the target's moments of inertia ratio, in the presence of initial deviations. The multiplicative state/measurement update procedures are developed for six-dimensional error dual quaternion to guarantee two normalization constraints in dual quaternions. The efficacy is demonstrated through numerical simulations and further validated with experimental trials on a dualmanipulator platform. The estimation performance outperforms a classic dual quaternion multiplicative extended Kalman filter, which ignores the spacecraft dynamics.