Control of Pendubot Using Input-output Feedback Linearization and Predictive Control
Abstract
In applying nonlinear model-predictive control to fast unstable systems, the main difficulty is that the optimization cannot be finished within one sampling interval. To solve this problem, a cascade-control scheme has been proposed, where input-output feedback linearization forms the inner loop and nonlinear predictive control the outer loop. Thus, the nonlinear predictive control has to handle only the internal dynamics which might be slow. For example, in the case of pendubot, an underactuated mechanical system with two rotary joints, the input-output dynamics are fast while the internal dynamics are slow. The proposed cascade approach was applied successfully to the experimental pendubot setup.