Abstract
This paper considers the shared control of a second order plant by a single machine and one person in the presence of exogenous disturbances. A mixed-initiative blending law that takes a passive measure of human intent as the input to a convex pair of exponential functions is proposed. The blending law is combined with a disturbance observer and homogeneous trajectory-tracking control law, which dynamically rescales to ensure that actuator magnitude constraints are respected. The globally uniform ultimate boundedness of the closed loop automatic controller is proved. The shared control system is tested in a simulation where a human user can impose a control input using a joystick. Even if large tracking errors are commanded by the user, the automatic controller is able to quickly return to tracking the desired path when the joystick is released.