Controllers with desired dynamic compensation and their implementation on a 6 DOF parallel manipulator

Investigates nonlinear controllers with dynamic compensation depending solely on the desired state and velocity computed from the position signal using a velocity observer. A feedforward controller and a controller similar to the computed torque controller are compared. The stability is proved, and adaptive versions of these two controllers are implemented on a 6 DOF parallel manipulator with uncoupled and highly nonlinear dynamics. Both controllers improved the tracking performance critically relative to a linear controller. We recommend using the feedforward controller, because it is simpler to implement and the difference of performance between both controllers is minor