Faculty of Engineering and Natural Sciences
Model Predictive Control: The Best is Yet to Come
Jan Maciejowski, Cambridge University
ABSTRACT: This talk advances the notion that Model Predictive Control (MPC) has the potential to make a huge technological and economic impact, comparable with those of the foundational concepts of Control. MPC is potentially an effective technology for the integrated control of various systems, ranging from standard "low-level" control tasks, such as attitude control and process regulation, to more "high-level" tasks such as mission-management, mid-mission re-planning, active fault-tolerance, and in general increased autonomy, of both vehicles and enterprises. MPC provides a unified approach to planning and control in dynamic environments. Its online, real-time optimisation can exploit knowledge of system models and behaviour constraints to make complex decisions without unnecessary conservatism. Furthermore, links to fundamental control theory enable MPC to provide guarantees such as assured levels of performance despite unknown disturbances. Basic MPC, for simple formation control for example, requires the online solution of a convex linear or quadratic program, which can be done very fast. However, for more complex problems such as guaranteed robustness or discrete-decision making, managing computational complexity becomes a key challenge to implementation. I will outline briefly some of the work we have done at Cambridge, aiming to address this and other challenges.
BIO: Jan Maciejowski is Professor of Control Engineering and Head of the Information Engineering Division at Cambridge University. He worked in the space industry before starting an academic career. He has researched in the areas of multivariable control, system identification, the application of control to macroeconomics, computer-aided control design, and predictive control. He is currently interested in applying predictive control ideas to autonomous systems. He is the author of 2 well-known textbooks, "Multivariable Feedback Design" (1989) and "Predictive Control with Constraints" (2002). In 2002 he was President of the Institute of Measurement and Control (UK), and from 2003 to 2005 he was President of the European Union Control Association. (More details available at:
October 12, 2009, 13:40, FENS L062