Kalman filtering for dynamic motion and model estimation
thesisposted on 23.05.2021, 12:57 authored by Randal Schumacher.
The fundamental task of a space vision system for rendezvous, capture, and servicing of satellites on-orbit is the real-time determination of the motion of the target vehicle as observed on-board a chaser vehicle. Augmenting the architecture to incorporate the highly regarded Kalman filtering technique can synthesize a system that is more capable, more efficient and more robust. A filter was designed and testing was conducted in an inertial environment and then in a more realistic relative motion orbital rendezvous scenario. The results indicate that a Dynamic Motion Filter based on extended Kalman filtering can provide the vision system routines with excellent initialization leading to faster convergence, reliable pose estimation at slower sampling rates, and the ability to estimate target position, velocity, orientation, angular velocity, and mass center location.