Characterization of command software for an autonomous attitude determination and control system for spacecraft

Abstract Space missions frequently carry equipment that must be accurately pointed toward remote targets. Therefore, effective attitude control is a vital part of almost every class of spacecraft. The component that governs the spacecraft’s rotational motion and pointing is the attitude determination and control system (ADCS). Due to the potentially unpredictable nature of some space missions, an ADCS that possesses adaptive capabilities will maximize the likelihood that the spacecraft remains effective throughout the mission timeframe. This paper presents an implementation of an adaptive ADCS that progressively learns the behavior of and adapts to changes in the spacecraft to more accurately control its attitude. Using various machine learning techniques, prototype software for an ADCS which is able to learn a nonlinear model of the spacecraft’s rotational dynamics has been developed. This software utilizes a database of previous maneuver information to predict maneuvers that will result in a desired set of sensor deltas. Attitude change maneuvers were tested and the results are presented.