See control theory
As with most engineering disciplines, automatic control has its roots in the solution of practical problems. This particular science stretches back millennia and advanced in four major 'bursts':
The Water Clocks of the Greeks and Arabs
Motivated by the need for the accurate determination of time, the Greek Ktesibios invented the float regulator around 270 BC for a water clock. This regulator would maintain the water level in a tank at a constant depth, which yielded a constant flow of water through a tube that would fill a second tank at a constant rate. Therefore the level of water in the second tank depended on the time elapsed. This was possible since as the water level dropped, a float would gradually open a valve, which would in turn replenish the reservoir--the simple mechanism that controls our flush toilets today!
Interestingly, not only did the ancient Greeks use the float regulator for accurate timekeeping, but they also used it for the automatic dispensing of wine (!), the design of siphons for maintaining constant water levels in tanks, and the opening of temple doors, among other things.
Between 800 and 1200 AD, various Arab engineers also designed many systems that relied on the principle of feedback control and introduced the concept of on/off control--a very simple but elegant method of regulation. Unfortunately, their fruitful research largely came to an end when Baghdad fell to the Mongols in 1258.
Eventually, the mechanical clock rendered the water clock and feedback control obsolete in the 14th century. Feedback control would not appear again until the Industrial Revolution.
The Industrial Revolution
During this time, "self-driven" machines such as advanced grain mills, boilers, furnaces, and steam engines were introduced which could not be adequately regulated by hand. This, of course, lead to the introduction of a wide variety of control devices including (once again) float regulators, temperature regulators, pressure regulators, and speed control devices (governors).
As these control systems grew increasingly complicated, trial-and-error design coupled with engineering intuition would begin to prove insufficient. In the mid 1800s, mathematics was first used to analyze the stability of feedback control systems. From differential equations to the Routh stability criterion to transfer functions, this mathematical modelling advanced rapidly up to the early 1900s, when the concept of systems theory emerged.
Mass Communication and the World Wars
Up to the 20th century, control system analysis had been performed purely in the time domain. Only in the 1920s and 1930s did the frequency domain mathematics of Laplace, Fourier, and Cauchy come into play at Bell Labs, where communications research was advancing very rapidly.
The main motivating problem in mass communication systems at the time was the need to periodically amplify the voice signals in long telephone lines without also amplifying the noise. It was in this area that H.S. Black demonstrated the usefulness of negative feedback in 1927. H. Nyquist and H.W. Bode then both introduced analysis techniques for control system design (Nyquist stability criterion, Bode plots) in the following decade. This link between control and communication would be further explored by Norbert Wiener, leading to his formulation of cybernetics.
During the World Wars, the development of feedback control became a matter of survival. It was key in the design of ship control and navigation systems, as well as in the accurate pointing of guns. Also, noise elimination took centre stage once again during the study of information processing problems associated with the newly invented radar. The work of the MIT Radiation Lab in this area would lead to the development and formalization of advanced control design techniques, which would prove crucial to the advancement of the field in the post-war years.
The Space and Computer Age
In the late 1950s, control design would surprisingly turn away from the frequency domain and return to the time domain differential equation techniques of the 1800s! Although this seems incredible at first glance, it made plenty of sense with the introduction of the complex nonlinear multivariable systems associated with aerospace applications. The frequency domain approach depended on linearity and time invariance, and worked best with single-input/single-output (SISO) systems since its graphical techniques were extremely inconvenient to apply to multiple inputs and outputs. These limitations would motivate the development of new, more powerful techniques based on linear algebra, leading to the introduction of modern control.
Around the same time with the rise of modern computing, digital control would also be introduced via the theory of sampled data systems. Discrete time systems--those modelled and controlled by computers--behaved quite differently, necessitating new techniques and new tools to deal with them. Despite the additional complications introduced by digital control, its advantages would prove to be tremendous, especially when the design of modern control systems became possible for the individual engineer with the introduction of the PC in 1983.
Perhaps I should have called this "a *brief* history of automatic control"? :)
REFERENCES:
First chapter of Lewis, F.L. Applied Optimal Control and Estimation.
Prentice-Hall, 1992.
http://www.theorem.net/theorem/lewis1.html
Franklin, G.F., J.D. Powell, A. Emami-Naeini. Feedback Control of
Dynamic Systems, 3rd ed. Addison-Wesley, 1994.
| 
