In-orbit Stereovision Tracking of RSOs

Experimental setup for validating stereovision tracking algorithms

Estimating the relative pose, structure and motion between cooperative spacecraft using on-board sensors is a challenging problem. When the spacecraft are non-cooperative, this challenge becomes far more complicated. This research attempts to solve this problem by developing a passive method for relative spacecraft pose and motion estimation using stereoscopic vision. Two cameras mounted on a microsatellite acquire images of feature points on a resident space object (RSO). The rotational and translational dynamics of the RSO relative to the microsatellite, as well as the RSO's orbital elements, are estimated using image sequences only. An estimation algorithm is developed based on a kinematically-coupled dynamical model and measurements of the projected feature points. The newly-proposed estimation process is performed in two stages: First, an algebraic solution based on stereovision tracking of feature points provides an approximation of all the relative states. This procedure results in an initialized state vector, which is fed into a nonlinear filter. The developed estimation algorithm reduces the sensitivity to initial uncertainties without linearizing the measurement equation or the process model.