Modern control systems increasingly consist of complex, interconnected components and must make reliable decisions in dynamic environments. PN 4 investigated how classical model-based control could be combined with data-driven and machine-learning methods without giving up rigorous guarantees for system behaviour.
The research followed the full systems and control design cycle – from developing data-integrated models to analysing system properties and designing controllers for networked systems. Applications in areas such as robotics and biology were used to test how these methods could remain both theoretically sound and practically applicable.
Research Focus
The research addressed five closely connected questions:
- Data-integrated models: How could first-principles models be combined with relationships learned from data?
- System guarantees: How could important system-theoretic properties be analysed and guaranteed in mixed model- and data-based systems?
- Controller design: How could model-based and data-driven approaches be combined to develop reliable controllers?
- Networked systems: How could interconnected systems be coordinated to support adaptive decisions at a higher system level?
- Applications: How could these methods be transferred to complex real-world systems, particularly in robotics and biology?
Projects
| Data-driven verification of system theoretic properties for nonlinear systems | Frank Allgöwer (David Remy) |
| Data-enhanced analysis and controller design for uncertain systems | Carsten Scherer |
| Learning from data - predictive control in adaptive multi-agent scenarios | Peter Eberhard (Frank Allgöwer) |
| Combining data and first principles in the control of muscle-redundant systems in Biomechanics and Robotics - exploring the contribution of prior knowledge through morphology variation | Syn Schmitt (Frank Allgöwer, David Remy) |
| Combining first principles and neural network models for interpretable, high-precision multi-step predictions (InMotion) | Steffen Staab (Daniel Weiskopf, Carsten Scherer) |
| Data-integrated low-dimensional surrogate models with guarantees | Benjamin Unger (Carsten Scherer) |
| Combining data and first principles in the synergy-based control of soft robotic systems | David Remy (Frank Allgöwer, Syn Schmitt) |
| Uncertainty quantification for indirect and direct data-driven control | Andrea Iannelli (Peter Eberhard) |
| Mixed first-principles and data-integrated modeling for control | Sebastian Trimpe (Carsten Scherer) |
| Determining system theoretic properties from input-output data | Frank Allgöwer (Syn Schmitt) |
| Optimization-based design of data-integrated controllers | Carsten Scherer (Sebastian Trimpe) |
| Theoretical guarantees for predictive control in adaptive multi-agent scenarios | Peter Eberhard (Frank Allgöwer) |
| Safe control of redundant musculoskeletal systems using nonlinear MPC | Syn Schmitt (Frank Allgöwer, Sebastian Trimpe) |
Project Network Coordinators
Frank Allgöwer
Prof. Dr.-Ing.Mechanical Engineering
Peter Eberhard
Prof. Dr.-Ing. Prof. E.h.Engineering and Computational Mechanics