The junior research group “Cyber-Physical Autonomous Systems” (CyPhy) aims at a new generation of reliable and trustworthy intelligent systems that seamlessly integrate with human environments. We unite the machine-centred and human-centred perspectives by modelling the entire human-AI interaction loop as dynamical systems. To drive these systems, our goal is to develop hybrid control strategies that combine established model-based (e.g., Model Predictive Control) and promising Machine Learning/data-driven (e.g., Deep Reinforcement Learning) approaches. We then apply these strategies to simulate, optimize, and implement intelligent systems with which humans can interact (more) naturally through new and more accessible user interfaces. With our application-driven research, we aim to bridge the gap between advanced theory and impactful, real-world solutions.

Projects

More about current and future projects of this group will follow.

Team

Lead

The junior research group leader, Dr. Arthur Fleig, joined ScaDS.AI Leipzig in March 2024. He studied Mathematics at the University of Bayreuth and completed his PhD on “Model Predictive Control of the Fokker-Planck equation” under Professor Lars Grüne’s supervision. Before joining ScaDS.AI, he worked as a postdoc in Professor Jörg Müller’s group in Bayreuth, where he applied his mathematical core expertise to the field of Human-Computer Interaction (HCI).

Since 2019, Fleig focuses on modelling, simulation, and optimal control of real-world-relevant dynamical systems. He develops and combines optimal control methods – such as Deep Reinforcement Learning and Model Predictive Control – with Human-Computer Interaction. In various international (UK, Norway, Finland) collaborative teams, he introduced the Optimal Feedback Control perspective to the HCI community. With OptiTrap, the first structural numerical approach to compute trap trajectories for acoustic levitation displays, he is one of a select group of HCI researchers to have published in the ACM Transactions on Graphics SIGGRAPH journal.

Fleig’s research passion lies in viewing interaction with intelligent systems from both a machine-centred and a human-centred perspective. This incorporates both the user-centred design approach as well as a simulation- and optimisation-based approach to modelling interaction with these systems. By considering all three elements of the human-AI interaction loop – the user, the interface, and the intelligent system – as dynamical systems, this group aims at a new generation of reliable and trustworthy intelligent systems with which humans can interact (more) naturally through new and more accessible user interfaces.

Team Members

Publications

• H. Selder, F. Fischer, P. O. Kristensson, and A. Fleig (2025): What Makes a Model Breathe? Understanding Reinforcement Learning Reward Function Design in Biomechanical User Simulation. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25). Article 592, 1–10. https://doi.org/10.1145/3706599.3719699
• V. Freiberger, A. Fleig, E. Buchmann (2025): “You don’t need a university degree to comprehend data protection this way”: LLM-Powered Interactive Privacy Policy Assessment. In Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’25), Article 36, 1–12. https://doi.org/10.1145/3706599.3719816
• A. Bauske and A. Fleig (2025): You Shall Not Pass: Warning Drivers of Unsafe Overtaking Maneuvers on Country Roads by Predicting Safe Sight Distance. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems (CHI ’25). Article 863, 1–23. https://doi.org/10.1145/3706598.3713768 [Honorable Mention Award, top 5%]
• V. Jüttner, A. Fleig, E. Buchmann (2025): ChatAnalysis Revisited: Can ChatGPT Undermine Privacy in Smart Homes with Data Analysis? i-com, vol. 24, no. 1, pp. 173-187. https://doi.org/10.1515/icom-2024-0072
• V. Freiberger, A. Fleig, E. Buchmann (2025): Explainable AI in Usable Privacy and Security: Challenges and Opportunities. Accepted at the Human-Centered Explainable AI (HCXAI) Workshop at CHI ‘25). 9 pp. https://arxiv.org/abs/2504.12931
• F. Fischer, A. Ikkala, M. Klar, A. Fleig, M. Bachinski, R. Murray-Smith, P. Hämäläinen, A. Oulasvirta, J. Müller (2024): SIM2VR: Towards Automated Biomechanical Testing in VR. In Proceedings of the 37th Annual ACM Symposium on User Interface Software and Technology (UIST), Article 79, 1–15. https://doi.org/10.1145/3654777.3676452
• V. Jüttner, A. Fleig, E. Buchmann (2024): ChatAnalysis: Can GPT-4 Undermine Privacy in Smart Homes with Data Analysis? Mensch und Computer 2024-Workshopband. Gesellschaft für Informatik e.V.. MCI-WS13: Workshop Mensch-Maschine-Interaktion in sicherheitskritischen Systemen. https://doi.org/10.18420/muc2024-mci-ws13-143