Saša V. Raković - Research
Dr. Raković's research expertise and activity span the broad areas of autonomy, controls, dynamics, systems, applied mathematics, optimization and set–valued analysis. Dr. Raković is best known for his research in model predictive control. His research has redefned and shaped modern model predictive control under uncertainty, resulting in global recognition of his work as the state of the art in robust model predictive control. His current research activity is driven by problems encountered in the felds of, or related to, smart autonomous and cyber–physical systems, which belong to the intersection of controls, dynamics, systems and optimization. Dr. Raković's current research efforts are focused on enabling transition of this uniquely versatile and powerful control methodology to smart autonomous and cyber–physical systems and, thus, facilitating its real–life and highly benefcial utilization for a wide range of important application areas including, but not exclusively limited to, smart transportation systems and autonomous vehicles, smart energy, water and heat systems, smart buildings and cities, smart manufacturing systems as well as smart ﬂow and supply chain networks.
The most important contributions of Dr. Raković's work fall within the felds of model predictive control and analysis of dynamics and control synthesis via optimization and set–valued methods. Dr. Raković's work in model predictive control has signifcantly contributed to theory, computation and implementation of conventional, robust and stochastic model predictive control. In fact, Dr. Raković is one of the global leaders in robust model predictive control, and one of the key pioneers of the tube model predictive control framework. Tube model predictive control has been recognized as a milestone contribution to, and a major paradigm shift in, model predictive control under uncertainty. In particular, his personal and collaborative research activities on model predictive control under uncertainty have set a foundational base for tube model predictive control, and they have introduced and defned its main generations.
- Rigid tube model predictive control.
- Homothetic tube model predictive control.
- Elastic tube model predictive control.
- Parameterized tube model predictive control.
Parameterized tube model predictive control has been referred to as an ingenious solution for robust model predictive control in a recent survey paper on model predictive control published in Automatica. Parameterized tube model predictive control outperforms rigid, homothetic and elastic tube model predictive control and other available methods for robust model predictive control, and it is the present state of the art in model predictive control under uncertainty. The parameterized tube model predictive control is, in a number of cases, equivalent to the dynamic programming based robust model predictive control and, thus, it can not be outperformed in these instances. Furthermore, in stark contrast to the exponential computational complexity of the dynamic programming based robust model predictive control, the parameterized tube model predictive control is implementatable due to its quadratic computational complexity.
Dr. Raković's most important work in analysis of dynamics and control synthesis via optimization and set–valued methods has dealt with previously long–standing problems. In particular, Zvi Artstein and Saša V. Raković have resolved important problems concerned with minimality of invariant sets and set invariance under output feedback. Dr. Raković has also significantly expanded classical results on the linear quadratic Lyapunov equations by developing theory and computations for Minkowski-Lyapunov equations, namely Lyapunov equations within the class of generic vector semi–norms.