We all have been accustomed to see a block diagram for a linear control system, where in the forward path, first sits the controller, then actuator, and then plant.

What if we flip the blocks of actuator and plant. I mean what if we place plant before actuator?

We know that, the closed-loop transfer function does not depend on the consecutive order of blocks in forward path, since: C(s)/R(s)=G(s)/(1+G(s)H(s)), where G(s) is the forward-path gain and H(s) is the feedback-path gain (sensor). The forward path gain could be decomposes as:

G(s)=Gcontroller(s)*Gactuator(s)*Gplant(s).

Therefore, since these transfer functions in forward-path are multiplied, then it does not differ what are their consecutive order.

Hence, let's flip the plant and actuator blocks in the forward path: G(s)=Gcontroller(s)*Gplant(s)*Gactuator(s).

What happens next? Nothing about the stability, but a great revamp of the required embedded system and controller design. Because instead of measuring plant out-put to send it into the feedback path, we only need to measure actuator out-put which is more convenient through the control engineering practice. In other words, we have mixed controller and plant blocks together, to establish a new controller which physically controls the actuator.

For stabilization, I think it works well. For example, assume an aircraft autopilot (controller). Through this revamped controller design, instead of measuring and regulating aircraft states as pitch-rate, we only measure and regulate actuator (elevator) deflection. If we can stabilize the elevator angular position to stay at zero, then we have stabilized the aircraft at a level-flight (stabilization), since zero elevator-deflection means level flight has been sustained.

For command following (not stabilization), we may set a mathematical trick as:

G(s)=Gcontroller(s)*Gactuator(s)*[Gplant(s)/Gactuator(s)]*Gactuator(s),

in this way, even for the case of command following (control) we still need to only measure actuator output.

- in the classical loop, we use actuator model as a software block in the controller simulation and design, while using plant as a hardware to be regulated by physical sensing of its output.

-in my revamped controller setup, we use plant model as a software block in the controller simulation and design, while using actuator as a hardware to be regulated by physical sensing of its output.

It seems physically sensing and regulating actuator dynamics, would be much convenient than physically sensing and regulating plant dynamics.