July 4, 2016 at 12:00 p.m. Dr. Ruth Milman, Assistant Professor at UOIT, will be presenting “The Application of Optimization to Model Predictive Control”.
Speaker: Dr. Ruth Milman
Assistant Professor – Department of Electrical, Computer and Software Engineering
Faculty Applied Science and Engineering, University of Ontario Institute of Technology
Day & Time: Monday, July 4, 2016
12:00 p.m. – 1:00 p.m.
Location: Room ENG 288
245 Church St., Toronto, ON, M5B 2K3
Contact: Dr. Maryam Davoudpour
Organizers: IEEE Women in Engineering (WIE), IEEE Magnetics Chapter, IEEE Instrumentation & Measurement/Robotics & Automation Joint Chapter and Computer Science Department Ryerson University
Abstract: Model predictive control (MPC) is the application of an optimal control scheme over a finite horizon. At each sample interval a cost function is minimized over a finite horizon and a resulting open loop controller is calculated. The control for the current sample interval is applied and the whole process is repeated at the next sample interval. By repeating the process at each sample interval, the resulting control scheme, which is technically open loop, inherits the benefits of a closed loop controller. These include some stability and robustness properties. By nature, MPC is computationally intensive and only makes sense when a there are constraints which must be enforced by the system. As would be expected, adding constraints into the system even further intensifies the computational requirements.
By nature, MPC is an optimal control strategy. If a true optimal control is computed when solving the minimization problem, then the solution is independent of the choice of the optimizer. It is only when time constraints force the need for suboptimal controls to be used that the actual algorithm plays a role in the quality of the resulting controller. Despite (or because of) this, the choice of optimization schemes plays a critical role in the real time application of MPC for a simple but important reason – the computational time it takes to solve for the optimal solution. MPC is a flexible framework which allows for control in the face of both linear or nonlinear systems, and can be applied to systems with either hard or soft constraints. How each problem is set up is critical to the choice of optimizer. These choices can drastically impact the computational effort which is required to solve for the resulting controller. As such, the choice and application of optimization schemes to MPC is of critical importance to the resulting performance of the systems.
Biography: Dr. Ruth Milman is an Assistant Professor in the Department of Electrical, Computer and Software Engineering with the Faculty of Applied Science and Engineering at the University of Ontario Institute of Technology. She has been with UOIT since June 2007, where she works in the Department of Electrical and Software Engineering, focusing in the field of control theory. Her research interests include optimization and computationally efficient algorithms for model predictive control as well as the application of both linear and nonlinear MPC to autonomous systems. She has worked on path planning for robotic applications in environments with both moving and stationary obstacles. She has worked extensively in the areas of nonlinear and optimal control theory and has developed algorithms for computation of the optimization problem that underlies Model Predictive control. Prior to coming to UOIT she did post-doctoral research at the University of Toronto from 2005 to 2007.
Ruth Milman obtained her PhD in 2004 from the Systems Control Group in the Department of Electrical and Computer Engineering at University of Toronto, Canada. Her dissertation focused on improving the speed and computational efficiency of a Linear Model Predictive Controller. As part of this she developed a novel algorithm for solving the quadratic programming subproblem in MPC. She obtained her MASC in 1997 from the Systems Control Group in the University of Toronto and her BASc (Honours) in Computer Engineering in 1995 from the Faculty of Applied Science and Engineering at the University of Toronto.