This work presents a survey of feedforward design. We will show why is important for motion systems the use feedforward, present an overview of different feedforward techniques (model inversion, command shaping, jerk derivative, spatial, multi input inverse dynamics and learning control).
In the second part of the report, inverse model techniques will be shown on three example systems in simulation. These systems are a single mass, mass-spring-damper , and a two mass system with flexible shaft. The first system will introduce to feedforward basis, the second to model inversion benefits in tracking error reduction and the last one is a commonly used system that will lay foundation to the real set up experiments.
With the real two-mass set up we will prove that when using feedback control in one of the masses movement and feedforward control in the other, the feedback signal deteriorates the performance of the mass controlled by feedforward.
At the end a technique will be presented to reduce both errors implementing a filter to create a special reference profile for the feedback control and another one for the feedforward controller.