Hybrid control systems: using formal verification to improve the control loop

The complexity of today’s technological applications means that automated and semi-automated processes have become proportionately more complicated. With consumers demanding more from automated services, the need for safety-critical and resilient systems - that is, systems capable of preserving stability despite uncertainties and recovering from contingencies - becomes more pressing. Complex engineering systems are continually changing, highly nonlinear, and combine continuous and discrete, smooth and abrupt dynamics. Their combined dynamics can be seen as a hybrid dynamical system. The process of checking in an automated way that a system behaves correctly is called ’formal verification’ in the theory of computer science; the elimination of negative behaviours falls into the field of control engineering.

But, how can we accurately test the performance of complex systems, and, if necessary, to modify their behaviour by means of a controller to meet desired specifications? This research will answer this question by applying methods beyond classical engineering procedures. It will integrate formal verification results into control schemes, which will lead to more effective designs. To this end, different disciplines will be combined, mainly, control engineering and theory, formal methods of computer science, and dynamical systems methods. Depending on the student’s interests, different application domains can be explored. This research would be part of the project DYVERSE (DYnamical-driven VERification of Systems with Energy considerations).