Timing-Robust Control over the Cloud Using On-Line Parametric Optimization

In this paper, we present a heuristic method for adapting a networked linear feedback controller to improve its robustness to timing complications, such as long delays, aborted computations, and dropped packets. The core concept of the approach is to log successful sampling and actuation events and then, at discrete time-points, use non-convex parametric optimization to improve the expected performance of the controller under the assumption that the future timing behavior will be similar to the current one. To reduce the time complexity of the optimization algorithm, automatic differentiation is integrated for Efficient gradient descent. The approach is evaluated on a physical ball and beam plant, where both the controller and optimization algorithm can be located in the Cloud.