Machine Learning Applications in Spacecraft State and Environment Estimation

Abstract:

There are some problems in spacecraft systems engineering with highly non-linear characteristics and noise where traditional nonlinear estimation techniques fail to yield accurate results. In this thesis, we consider approaching two such problems using kernel methods in machine learning – orbit determination and spacecraft magnetic field environment estimation.

We present a novel formulation and solution to orbit determination of spacecraft and spacecraft groups which can be applied in very weakly observable and highly noisy scenarios. We show that a novel machine learning approach of mixture distribution regression can learn to estimate orbits of a group of satellites. We provide asymptotic convergence conditions for the approach. We present a ground station network architecture that can perform orbit determination using Doppler-only observations over the network.

We present a machine learning solution to the spacecraft magnetic field interference cancellation problem using distributed magnetometers paving the way for space magnetometry with boom-less CubeSats. The proposed approach models this as simple regret minimization in contextual bandits. We validate our solutions using datasets from on-orbit spacecraft. These contributions will help the growing number