Principles of feedback control

Abstract

In this chapter, fundamental properties of feedback control are discussed with engineering applications in mind. However, feedback appears in many other disciplines, such as agriculture, biology, medicine, pharmaceutics, economics, business management, political science, etc. A typical example of feedback control is steering of an automobile: we would like to keep the vehicle on the road (in the cruising lane). The road curvature determines the desired path to be followed. On the other hand, unpredictable gusting winds, bumps, or potholes on the road may move the vehicle off the cruising lane, unless corrective steering action is taken. As the driver (or autopilot) detects, using vision or other sensors, deviation from the desired path he/she can take the corrective steering action. Clearly, without the feedback from vision (or other sensory mechanism), the automobile cannot be kept on the road for a long time, even if we know the desired path a priori. In this example, the position of the vehicle relative to the center of the lane is the system output to be controlled, vision or another sensor provide a measured value of the output to the controller (human driver or automatic steering mechanism). This information is processed together with the desired path to be followed (i.e., reference input), and then steering angle (i.e., the control input) is determined by the controller. The immeasurable variables such as wind, bumps, potholes, as well as measurement errors due to imprecise sensing, represent the disturbances in this feedback system. The main goal of feedback control is to reduce the effect of uncertainty (i.e., disturbances in the above example) on the output of the system. In fact, it is widely accepted by control engineers that the only reason to use feedback is to cope with uncertainty.