Publications
SySCoRe: Synthesis via Stochastic Coupling Relations
Oliver Schön, Birgit van Huijgevoort, Sofie Haesaert, Sadegh Soudjani
IEEE 61st Conference on Decision and Control (CDC 2022).
Establishing formal guarantees while allowing for epistemic uncertainty is still a great challenge. We investigate ways of relating parametric stochastic systems to their nominal model.
With Alessandro Abate, Henk Blom, Joanna Delicaris, Arnd Hartmanns, Abolfazl Lavaei, Hao Ma, Mathis Niehage, Anne Remke, Stefan Schupp, Lisa Willemsen. 9th International Workshop on Applied Verification of Continuous and Hybrid Systems (ARCH22).
A friendly competition for formal verification and policy synthesis of stochastic models showcasing the performance and limits of state-of-the-art solution approaches.
26th ACM International Conference on Hybrid Systems: Computation and Control (HSCC 2023).
Developed MATLAB toolbox for synthesizing robust controllers for stochastic continuous-state systems to satisfy temporal logic specifications and quantifying the associated formal robustness guarantees.
This paper addresses the problem of data-driven computation of controllers that are correct by design for safety-critical systems and can provably satisfy (complex) functional requirements. With focus on continuous-state uncertain systems , we propose a two-stage approach that decomposes the problem into a learning stage and a robust formal controller synthesis stage. The first stage utilizes available Bayesian regression results to compute robust credible sets for the true parameters of the system. For the second stage, we introduce methods for systems subject to both stochastic and parametric uncertainties. We provide for the first time simulation relations for enabling correct-by-design control refinement that are founded on coupling uncertainties of stochastic systems via sub-probability measures. The presented relations are essential for constructing abstract models that are related to not only one model but to a set of parameterized models. The results are demonstrated on a linear model and the nonlinear model of the Van der Pol Oscillator.
26th ACM International Conference on Hybrid Systems: Computation and Control (HSCC 2023).
In this ongoing work, we address data-driven computation of controllers that are correct by design for safety-critical systems and can provably satisfy complex functional requirements. We propose a two-stage approach that decomposes the problem into a data-driven stage and a robust formal controller synthesis stage. The first stage utilizes available Bayesian linear regression methods to compute robust confidence sets for the true parameters of the system. The second stage develops methods for systems subject to both stochastic and parametric uncertainties. We provide simulation relations for enabling control refinement that are founded on coupling uncertainties of stochastic systems via sub-probability measures. Such relations are essential for constructing abstract models that are related to not only one model but to a set of parametric models.
Accepted to IEEE 62st Conference on Decision and Control (CDC 2023).
We address two limitations of existing approaches for formal synthesis of controllers for networks of stochastic systems satisfying complex temporal specifications. Firstly, whilst existing approaches rely on the stochasticity to be Gaussian, the heterogeneous nature of composed systems typically yields a more complex stochastic behavior. Secondly, exact models of the systems involved are generally not available or difficult to acquire. We design controllers based on parameter uncertainty sets identified from observed data and approximate possibly arbitrary noise distributions using Gaussian mixture models whilst quantifying the incurred stochastic coupling.
Alessandro Abate, Henk Blom, Nathalie Cauchi, Joanna Delicaris, Sofie Haesaert, Birgit van Huijgevoort, Abolfazl Lavaei, Anne Remke, Oliver Schön, Stefan Schupp, Fedor Shmarov, Sadegh Soudjani, Lisa Willemsen and Paolo Zuliani
Open challenges are extension to LTL specifications and scalability.
Multi-Objective Physics-Guided Recurrent Neural Networks for Identifying Non-Autonomous Dynamical Systems
Oliver Schön, Ricarda Samantha-Götte, Julia Timmermann
14th IFAC Workshop on Adaptive and Learning Control Systems (ALCOS 2022).
Learning dynamics models from data whilst utilizing physics-based knowledge for efficient convergence and physically plausible models. Combining a multi-objective approach, an approximate dynamics model, and a recurrent neural network.
All Publications on Google Scholar
What I have been working on
Other Projects
Coming soon
Meet My Collaborators

Dr. Sadegh Soudjani
Newcastle University, United Kingdom

Dr. Sofie Haesaert
TU Eindhoven, The Netherlands

Birgit van Huijgevoort
TU Eindhoven, The Netherlands