My vision is to advance fundamental understanding of multi-agent decision-making in uncertain and dynamic environments. Towards this vision, I develop game theory, distributed control, stochastic and data-driven control. I address applications ranging from air traffic control and rescue/disaster response robotics to power grid systems. The following presentation summarises my research. ETH Inaugural Lecture, May 10, 2017.
Selected recent publications:
I. Usmanova, A. Krause and M. Kamgarpour, Safe convex learning under uncertain constraints, AISTATS - International Conference on Artificial Intelligence and Statistics, Naha, Okinawa, Japan, 2019, to appear
T. Tatarenko and M. Kamgarpour, Learning Generalized Nash Equilibria in a Class of Convex Games, IEEE Transactions on Automatic Control, accepted, 2018
P. G. Sessa, D. Frick, T. A. Wood and M. Kamgarpour, From Uncertainty Data to Robust Policies for Temporal Logic Planning, Hybrid Systems: Computation and Control, ACM Lecture Notes in Computer Science, to appear, 2018
O. Karaca, P. G. Sessa, N. Walton and M. Kamgarpour, Designing Coalition-Proof Mechanisms for Auctions over Continuous Goods, IEEE Transactions on Automatic Control, 2019, to appear
L. Furieri and M. Kamgarpour, The Value of Communication in Synthesizing Controllers given an Information Structure, IEEE Transactions on Automatic Control, 2019, to appear
Formal methods and control: I work on developing provably safe feedback control policies that can adapt to an underspecified changing environment and achieve complex tasks, such as search and rescue and disaster response. To formulate this problem, I develop probabilistic models of the uncertainties in the environment, and define a formal language framework, such as temporal logic, to capture the complex specifications. The resulting control problem can then be cast as a mixed integer stochastic multi-stage optimisation problem. To solve this problem, I advance stochastic optimization and online safe learning. I currently lead an SNSF project on this topic with two doctoral students.Power grid system: I advance optimisation and control algorithms towards achieving a stable and secure electrical power grid despite increased uncertainties from the renewable energy sources and the liberalised electricity markets. The optimisation and control problems in a power system often involve large interconnected electricity networks and multiple interested stakeholders. Hence, they can best be captured using a multi-agent decision-making framework. My ERC project addresses this topic. Its first focus is on developing fundamental understanding of open challenges in distributed control and game theory. Second, equipped with this understanding, we develop scalable approaches to the large-scale control problems and we design mechanisms to coordinate decision-making of interconnected agents.
Air traffic system: My focus is on developing optimal control algorithms for safe and fuel-efficient aircraft trajectory synthesis, taking into account meterological uncertainties such as storms affecting aircraft routes. With my collaborators in Spain, we have developed stochastic models of meteorological forecast data to characterize safe airspace regions and to optimise aircraft trajectories accordingly. A challenging aspect of this control problem is ensuring the developed algorithms are compatible with air traffic procedures so that they have high levels of trust and acceptance by air traffic controllers. To address this issue, we collaborate with air traffic controllers and airlines.
Current PhD students: Orcun Karaca, Luca Furieri, Ilnura Usmanova, Pier Giuseppe Sessa, Yimeng Lu
PhD applicants: I welcome applications from motivated students with a strong background and interest in mathematics and control theory. If you are interested in a PhD position, please contact me with your undergraduate and Master's transcripts, CV, a brief statement of your research interests and goals, and contact information of three references.
Semester and Masters' students applicants: Please browse our institute's website for a list of open projects. You are also welcome to contact me if you have interests aligned with the context of my research discussed above or my publications. Your application should include your transcript and CV as well as a paragraph on the problems you are interested to work on.
I currently teach two Master's level courses: Linear System Theory and Game Theory. Other courses I have helped in teaching are: Hybrid Systems (UC Berkeley, Electrical Engineering), Nonlinear Control (UC Berkeley, Electrical Engineering and Mechanical Engineering), Intermediate Dynamics (UC Berkeley Mechanical Engineering), Calculus II (UC Berkeley, Mathematics).