Vision-Based Relative Navigation Using Dual Quaternion for Spacecraft Proximity Operations

Abstract

Dual quaternion is the parameter that can describe position and attitude in a unified form. It is extension of dual number and quaternion; so, mathematical and parameter characteristics of those are holding. The attitude and orbital kinematics model can be concurrently integrated in a simplified and intuitive formulation with dual quaternion. From these merits, dual quaternion-based relative navigation for spacecraft proximity operation is investigated in this paper. Especially, the case is considered which interested points are not coincident with center of mass of spacecraft. At this time, it is obvious that position is affected by rotational motion of rigid body. With conventional decoupled relative kinematics, this attitude-position coupling effect cannot be described at once. However, the new dual quaternion-based relative kinematics can represent coupled orbital motion in a simple form. Motivated by these backgrounds, in this paper, to estimate relative position and attitude states using vision sensor, dual quaternion-based relative kinematics is formulated. To reduce the number of state variables, error dual quaternion is applied, which makes the number of state variable be six instead of eight, by using parameter constraints of dual quaternion. Extended Kalman filter and unscented Kalman filter are adopted to realize relative navigation. Since the necessity of velocity information in dual quaternion kinematics, two types system models are constructed-velocity propagation and velocity measurement model. In addition, the six line-of-sight measurements are used to update the dual quaternion parameter. Simulation results show that the kinematics model is accurately derived, states are well estimated, and statistics of estimation errors are consistent with estimated error covariance.