Linearizing Feedforward/Feedback Attitude Control

This paper introduces a fundamentally new approach to attitude control theory in which a linear form is postulated for the closed-loop rotational error dynamics, then the exact control law required to realise it is derived. The nonminimal (four-component) quaternion form is used for attitude because it is obglobally nonsingular, but the minimal (three-component) quaternion form is used for attitude error because it has no nonlinear constraints to prevent the rotational error dynamics from being linearised, and the definition of the attitude error is based on quaternion algebra. This approach produces the first attitude control law that linearises the closed-loop rotational error dynamics exactly, without any ittitude singularities, even if the control errors become large. The control errors might become large if the attitude commands are physically unachievable, for example, or if large disturbances occur. Therefore we do not have to assure that the commands are always physically achievable, or the disturbances always small, to guarantee stability or to realize linearity.