May 29, 2015, Webb 1100
UC Santa Cruz, Applied Mathematics and Statistics
Advances in computational mathematics have made it possible to solve complex nonlinear optimal control problems. This talk focuses on Pseudospectral (PS) computational optimal control methods that have moved rapidly from mathematical theory to real-world applications. A unified PS framework based on arbitrary grids will be presented. Such a unified viewpoint provides a way to compare performance among different PS methods and suggests guidelines for choosing proper grids and discretization approaches for solving constrained nonlinear optimal control problems. Recent advances on computational optimal control for systems with parameter uncertainty will also be presented. Applications of such algorithms on optimal search and ensemble control problems will be discussed.