On the approximation of constrained infinite-horizon linear quadratic regulator problems
Event details
| Date | 29.06.2017 |
| Hour | 10:15 › 11:15 |
| Speaker | Michael Mühlebach, Institute for Dynamic Systems and Control, ETH |
| Location | |
| Category | Conferences - Seminars |
Abstract: I will construct approximations to constrained infinite-horizon linear quadratic optimal control problems arising in model predictive control.
Input and state trajectories will be parameterized as a linear combination of basis functions and the dynamics are approximated by a Galerkin
approach. I will argue that the resulting optimization problems lead to inherent closed-loop stability and recursive feasibility
when applied in the context of model predictive control. Moreover, I will derive a bound on the suboptimality of the approximation and present
simulation results indicating that the underlying infinite-dimensional optimal control problem is well-approximated. In addition, I will show
experimental results to confirm that the approach works in practice.
Bio: Michael Mühlebach received the Bachelor and Master degrees from ETH Zurich in 2010 and 2013, respectively. He received the Outstanding D-MAVT Bachelor Award for his Bachelor studies and the Willi-Studer prize for the best Master degree in Robotics, Systems, and Control. He did his master thesis on variational integrators for Hamiltonian systems and their application to multibody dynamics. His main research interests include multibody dynamics, the control of nonlinear systems, and model predictive control.
Input and state trajectories will be parameterized as a linear combination of basis functions and the dynamics are approximated by a Galerkin
approach. I will argue that the resulting optimization problems lead to inherent closed-loop stability and recursive feasibility
when applied in the context of model predictive control. Moreover, I will derive a bound on the suboptimality of the approximation and present
simulation results indicating that the underlying infinite-dimensional optimal control problem is well-approximated. In addition, I will show
experimental results to confirm that the approach works in practice.
Bio: Michael Mühlebach received the Bachelor and Master degrees from ETH Zurich in 2010 and 2013, respectively. He received the Outstanding D-MAVT Bachelor Award for his Bachelor studies and the Willi-Studer prize for the best Master degree in Robotics, Systems, and Control. He did his master thesis on variational integrators for Hamiltonian systems and their application to multibody dynamics. His main research interests include multibody dynamics, the control of nonlinear systems, and model predictive control.
Practical information
- General public
- Free