 |
Center for Nonlinear Studies |
We consider control systems where the feedback information has special characteristics such as quantization and faulty measurements. These characteristics may be the result of using low cost sensors and limited bandwidth channels. Using feedback with these characteristics we seek to achieve stability while addressing the adverse conditions which arise in most control systems. These conditions include external disturbances, model uncertainties and unknowns, and delays. We start by showing how to achieve the nonlinear Input to State Stability and disturbance rejection using dynamic quantization. We also show that this stability is maintained under small model uncertainties and delays. We then show under which conditions robustness to faulty measurements can be achieved using tools from Compressed Sensing, while maintaining stability to measurement noise. We finish by presenting an application, automatic landing controller using a camera for feedback, in which the feedback has these characteristics. We show that addressing these special characteristics in this application is required in order to achieve the desired performance.
Host: Misha Chertkov