The "sense-and-correct" nature of feedback controllers make them an appealing choice for systems whose actuators, or environments, are highly variable. If the system also requires high performance (e.g. an industrial robot, a car, or an aircraft), the usual approach is to use a state-space feedback controller derived from a physics-based model. And when performance is less critical (e.g. for toys and appliances), the traditional choice has been to tune a low-cost proportional-derivative-integral (PID) controller.
Over the last few years, much has changed. The dramatic decline in the cost of accurate sensors and fast microcontrollers have made state-space controllers practical even for inexpensive toys. In addition, modeling approaches have become far more reliant on measurement and computation rather than physics and analysis. In this course, we examine the theory and application of this arc of alternatives to control, starting with PID, then moving to physical-modeling and state-space, and ending with state-space using measurement-based modeling. In each case, you will design and test controllers with your own copter-levitated arm, to solidify your understanding and to gain insight in to the practical issues.
PLEASE NOTE: This is intended to be an advanced course and students should have a background in linear algebra and differential equations, as well as some experience with control systems. IN ADDITION: THIS IS A BETA COURSE, THINGS WILL GO WRONG. We are testing a new type of on-line class, one where students use advanced concepts to design and then examine performance results on their own hardware. There will be difficulties, and we will be updating content and focus in response to student input.
