Dr. Andrea Martinelli

My research lies at the intersection of optimal control, reinforcement learning, and decentralized control methods. In terms of optimal control, my main focus has been on improving scalability and feasibility issues of the linear programming approach to solve dynamic programming problems. When the dynamics of the systems is not known, one can learn an (optimal) controller from data on systems behaviour over time. In case a central control unit cannot be implemented, one can design decentralized control strategies based on local information or data. We develop such decentralized strategies for discrete-time nonlinear systems, based on the concept of dissipativity.