Improving the reliability of safety-critical systems

Airplanes, cars, and power systems are examples of safety-critical control systems, whose reliable and autonomous functioning is critical. It is of interest to design controllers for these systems that provably satisfy desired behavior, since malfunctions have disastrous consequences. Even though it is not always possible to completely avoid malfunctions, for example, due to uncertain influences, it is crucial to minimize the likelihood of these events. PhD candidate Birgit van Huijgevoort looked into improving the reliability of safety-critical systems by giving guarantees on their functional behavior.

To this end, the researcher explored the area of automatic, correct-by-design control synthesis for systems evolving over a continuous state space and develop formal methods to synthesize provably correct controllers subject to temporal logic requirements.

Manage uncertain factors

Multiple uncertain factors, such as stochastic influences on the dynamics, play a huge role in the behavior of safety-critical systems. When ignoring these uncertain influences, it is impossible to achieve guarantees that actually lead to reliable and safe behavior. That is why it is crucial to manage those uncertain factors and to obtain probabilistic guarantees on the behavior of safety-critical systems.

Guarantees on behavior autonomous systems

Van Huijgevoort developed formal methods that automatically synthesize controllers, while taking multiple uncertain influences into account. “If we keep on improving these methods, we will be able to give accurate guarantees on the behavior of autonomous systems in safety-critical situations. By doing so, we do not only make these technological advances possible but also reliable”, the PhD researcher says.

 Title of PhD-thesis: Automatic Control Synthesis with Temporal Logic Requirements: Stochastic, uncertain, and nonlinear systems. Supervisors: Sofie Haesaert, and Siep Weiland.