The following is what we believe the audience will take away after attending this workshop.

*A.* *Thinking Differently about Controls*

Beyond introducing a new concept or technique of controls, this workshop sets to stimulate the mind and encourage both creative and critical thinking, starting with the very concepts of model and disturbance.

All methods of control design are based on a nominal model of the plant, be it mathematical (differential or state equation, transfer function), empirical (heuristics, expert system, fuzzy logic), statistical (data-driven), or things like artificial neural network that mimics the plant dynamics. And the word disturbance usually refers to some external force acting on the plant. To the audience it will become soon evident that ADRC doesn’t rely on a model in its nominal sense. Rather, in ADRC, the controller is designed based on a simple, idealistic model (let’s call it design model), such as the frictionless, rigid body movement in servo design, which can be described as a double-integrator. The difference between the actual plant and the design model, however, is now deemed as “disturbance” to be rejected.

As such, the concept of “disturbance”, denoted variously to avoid confusion as total disturbance, total uncertainty, or generalized disturbance, is quite different from its usual meaning, and it comes to represent both the internal dynamics and the external forces, all of which could be unknown. The central tenant of ADRC is that if this total disturbance can be measured, or, more preferably, estimated, then it can be cancelled and the control becomes “active”, forcing the plant to behave like the design model. It is here the novelty of ADRC becomes evident: instead of using the means of feedback to change system dynamics, ADRC advocates a different venue of design: using the technique of “disturbance” estimation and cancellation, the uncertainties, both internal and external, are forcefully removed, making control design, whether it’s regulation or tracking, a much simpler task, without any hardware change.

*B.* *Originality and Creative Thinking*

The above discussion may, and probably does, sound like a fantasy, which may explain why it took so long for the idea to take hold. In fact, the lack of mathematical proof and persuasion has been cited as the main reason. But the lesson here is that not all original and creative ideas were born with rigor through logical deduction, and that, if history tells us anything, control is fundamentally an experimental science [49] where inventions first took place, long before any theoretical validation follow the suit. This is the case with the invention of the fly ball governor, this is the case with the control of the first powered aircraft of the Wright Brothers, and this is definitely the case with ADRC. In engineering, originality and creativity have the habit of leading, not following, the mathematical rigor and the theoretical development, which in turn justify and explain the inventions. Much progress has been made in the theoretical analysis of ADRC, as will be introduced in the workshop and it was stimulated the engineering ingenuity. It is this kind of symbiotic relationship between practice and theory that will keep the field of control engineering vibrant.

*C.* *Centrality of Disturbance Rejection*

Using the term disturbance in its wider sense discussed above, disturbance rejection is at the very root of control engineering, and it transcends artificial boundaries of theoretical investigations, such as those characterized as adaptive, robust, optimal, nonlinear, etc. In addition, dealing with disturbance is fundamental to all field problems, such as those in wind turbine, hydro and solar power generation, DC-DC converter, regulations of flow, pressure, temperature and web tension etc., computer hard disk drives, to name just a few. As such, disturbance rejection has become a unifying theme across the fields of academic studies and engineering practice, with the quality of disturbance rejection used as a measuring stick for the quality of control design. In disturbance rejection we could perhaps find the long awaited bridge between theory and practice so that our boundless energy and brainpower in academic research could be channeled to resolve challenging problems in the fields of control engineering practice, which in turn provides the stimulus for further research.

*D.* *Disturbance Rejection: Active vs. Passive*

As widely useful and well-developed as it is, feedback control by itself, from the fly ball governors in 1780s to the PIDs of today, can be seen as largely passive in disturbance rejection, for it waits for the disturbance to run its course and induce an output deviation from the set point before any action is taken. That is, feedback control reacts to such deviation, in order to eliminate it. To keep the deviation small, the controller must be made sensitive to small errors. That is, the control loop must have a high gain or high bandwidth that usually comes with high costs. To seek economy, engineers search for answers elsewhere. This is where active disturbance rejection comes in: used in conjunction with feedback control, it can make the deviation small at much lower cost, by not letting the disturbance run its course. The idea is quite intuitive: through measurement or estimation the disturbance information is obtained, based on which a cancelling force is applied accordingly, nulling its effect before a significant deviation occurs at the output. This is why energy saving is commonly associated with ADRC: energy is wasted in both the output deviation and its correction. Thus, making the deviation small saves energy both ways. ADRC provides a conceptual framework to understand such nuances and insight in automatic control and to investigate various techniques of active control.

*E.* *The Challenge of Field Applications*

The concept of ADRC is quite intuitive but the application of it could be deceivingly challenging. Great progress can be made when the application engineer, with the full understanding of the physical process, is able to pinpoint the nature of the disturbing forces. Only after such characterization of the disturbance could the engineer begin to devise a way to obtain, through sensing or numerical means, the disturbance information. That is, the application engineer needs to fully understand the causal relationship between the disturbance and the output deviation, based on which the controller is made to actively mitigates the cause before the effect takes place. This is the secret, or lesson, of ADRC.