Skip to main content

Learning to Control - Smart and Data-Driven Formal Methods for Cyber-Physical Systems control

Periodic Reporting for period 1 - L2C (Learning to Control - Smart and Data-Driven Formal Methods for Cyber-Physical Systems control)

Reporting period: 2020-09-01 to 2022-02-28

L2C is a pluridisciplinary project at the frontier between Control Engineering, Computer Science
and Applied Mathematics.Its goal is to develop automatic control techniques for the emerging Cyber-Physical systems (think of self-driving cars, smart grids, or coordinated moving robots), for which classical automation techniques turn out to be inefficient.

L2C impacts both fundamental Science and Engineering, as the theoretical
research is driven and fostered by cutting edge technological challenges.
We have gathered recent fundamental advances in the theory of control, and showed that it may be advantageously combined into a state-of-the-art, off-the-shelf, automatic controller design methodology, that can tackle complex systems, and even data-driven systems.
Until now, our advances are mainly theoretical, but we also have developed a couple of proofs-of-concept showing that we are in the good direction.
We have developed, or generalized, symbolic control, and data-driven control techniques, whose goal is to automatically control an arbitrarily complex control system, by representing it with 'symbolic tools', which is a representation that a computer can naturally handle. We have both developed new techniques, and generalized classical techniques to complex systems. We will now put all these techniques under a common umbrella, and implement this toolbox in a common software solution, which will be seamlessly usable by engineers, without having to rely on complex theories from systems and control.